Machine_Learning_Fairness.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "ETt1u81t4pAR"
   },
   "source": [
    "## **Machine Learning Final Assessment**\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "MQIS-QkbAobp"
   },
   "outputs": [],
   "source": [
    "#importing the aif360 library\n",
    "\n",
    "!pip install aif360"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "I6NveH0lR1ha"
   },
   "outputs": [],
   "source": [
    "#importing the fairlearn library\n",
    "#prevents errors when importing \n",
    "\n",
    "! pip install fairlearn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "YwTXop7mAt2o",
    "outputId": "da8e9bf2-9610-4b5d-8eb2-dfaeacd6a658"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.7/dist-packages/aif360/data/raw/adult\n"
     ]
    }
   ],
   "source": [
    "cd\t/usr/local/lib/python3.7/dist-packages/aif360/data/raw/adult"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "xNabbw60Ylw2"
   },
   "outputs": [],
   "source": [
    "#adult dataset import \n",
    "\n",
    "!wget https://archive.ics.uci.edu/ml/machine-learning-databases/adult/adult.data\n",
    "!wget https://archive.ics.uci.edu/ml/machine-learning-databases/adult/adult.test\n",
    "!wget\thttps://archive.ics.uci.edu/ml/machine-learning-databases/adult/adult.names"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "lXfk-iDbsKIb",
    "outputId": "7539ce24-901b-40be-ba38-3199f196fbd8"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.7/dist-packages/aif360/data/raw/german\n"
     ]
    }
   ],
   "source": [
    "cd /usr/local/lib/python3.7/dist-packages/aif360/data/raw/german"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "GEvXcGvUAupR"
   },
   "outputs": [],
   "source": [
    "#german dataset import\n",
    "\n",
    "!wget\thttps://archive.ics.uci.edu/ml/machine-learning-databases/statlog/german/german.data\n",
    "!wget\thttps://archive.ics.uci.edu/ml/machine-learning-databases/statlog/german/german.doc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "id": "RNYXO_hBvQx3"
   },
   "outputs": [],
   "source": [
    "#imports for the assignment\n",
    "\n",
    "from matplotlib import pyplot as plt\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import os\n",
    "import glob\n",
    "import random\n",
    "from numpy.random import RandomState\n",
    "import sklearn\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "import pdb\n",
    "\n",
    "from IPython.display import display\n",
    "\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.cluster import KMeans \n",
    "from sklearn.decomposition import PCA\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.model_selection import *\n",
    "from sklearn.model_selection import cross_val_score, GridSearchCV, train_test_split\n",
    "from sklearn.metrics import accuracy_score, f1_score\n",
    "\n",
    "from aif360.datasets import BinaryLabelDataset\n",
    "from aif360.datasets import AdultDataset, GermanDataset \n",
    "from aif360.sklearn.datasets import fetch_adult\n",
    "from aif360.sklearn.metrics import disparate_impact_ratio, average_odds_error, generalized_fpr, equal_opportunity_difference\n",
    "\n",
    "from aif360.algorithms.preprocessing.reweighing import Reweighing\n",
    "from aif360.algorithms.preprocessing.optim_preproc_helpers.data_preproc_functions import load_preproc_data_adult, load_preproc_data_german\n",
    "from aif360.metrics import BinaryLabelDatasetMetric, ClassificationMetric\n",
    "\n",
    "#setting random seed as the beginning of my candidate number\n",
    "np.random.seed(244)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "4DlFLngL9An9"
   },
   "source": [
    "## Splitting into Test and Train"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "OOIKi6uITAGH"
   },
   "source": [
    "### Functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "id": "g69Yytw9Qcnm"
   },
   "outputs": [],
   "source": [
    "#function to split and normalise each dataset\n",
    "\n",
    "def splitdata (dataset):\n",
    "\n",
    "  #sensitive attributes\n",
    "  privileged_groups = [{'sex': 1}]\n",
    "  unprivileged_groups = [{'sex': 0}]\n",
    "  dataset_orig = dataset(['sex'])\n",
    "\n",
    "  #splitting into test and train\n",
    "  train, test = dataset_orig.split([0.7], shuffle=True)\n",
    "  print(\"training data size\", train.features.shape)\n",
    "  print(\"dataset feature names\", train.feature_names)\n",
    "\n",
    "  #normalising the dataset\n",
    "  scale_orig = StandardScaler()\n",
    "  X_train = scale_orig.fit_transform(train.features)\n",
    "  y_train = train.labels.ravel()\n",
    "\n",
    "  X_test = scale_orig.transform(test.features) \n",
    "  y_test = test.labels.ravel()\n",
    "\n",
    "  #getting to know the dataset\n",
    "  #exploration step\n",
    "\n",
    "  #shapes of each dataset\n",
    "  print('The whole adult dataset shape is ' + str(dataset_orig.features.shape))\n",
    "  print('The training adult dataset shape is ' + str(train.features.shape))\n",
    "  print('The testing adult dataset shape is ' + str(test.features.shape))\n",
    "\n",
    "  #printing the attributes and labels\n",
    "  print('Favourable label is ' + str(train.favorable_label) +' and unfavourable label is '+ str(train.unfavorable_label))\n",
    "  print('Protected attribute names ' + str(train.protected_attribute_names))\n",
    "  print('Privileged protected attribute '+ str(train.privileged_protected_attributes) +' and unpriviledged protected attribute ' + str( \n",
    "      train.unprivileged_protected_attributes))\n",
    "  print('Dataset feature names '+ str(train.feature_names))\n",
    "\n",
    "  #returning dataset specific splits and labels\n",
    "  return train, test, X_train, y_train, X_test, y_test, dataset_orig, privileged_groups, unprivileged_groups"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "gm5CcqoqCq3x"
   },
   "source": [
    "### Adult"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "Fq2HDkmPpeed"
   },
   "source": [
    "Will use Logistic Regression Model and use the training data to perfrom 5-fold cross validation. Using a hyperparameter accuracy-generalisation trade off. Logistic regression classifier is a probablistic classfier \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "3_EuYAdsRYQc",
    "outputId": "595e0f05-f559-40c6-afa5-b2d6f5adc63a"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "training data size (34189, 18)\n",
      "dataset feature names ['race', 'sex', 'Age (decade)=10', 'Age (decade)=20', 'Age (decade)=30', 'Age (decade)=40', 'Age (decade)=50', 'Age (decade)=60', 'Age (decade)=>=70', 'Education Years=6', 'Education Years=7', 'Education Years=8', 'Education Years=9', 'Education Years=10', 'Education Years=11', 'Education Years=12', 'Education Years=<6', 'Education Years=>12']\n",
      "The whole adult dataset shape is (48842, 18)\n",
      "The training adult dataset shape is (34189, 18)\n",
      "The testing adult dataset shape is (14653, 18)\n",
      "Favourable label is 1.0 and unfavourable label is 0.0\n",
      "Protected attribute names ['sex']\n",
      "Privileged protected attribute [array([1.])] and unpriviledged protected attribute [array([0.])]\n",
      "Dataset feature names ['race', 'sex', 'Age (decade)=10', 'Age (decade)=20', 'Age (decade)=30', 'Age (decade)=40', 'Age (decade)=50', 'Age (decade)=60', 'Age (decade)=>=70', 'Education Years=6', 'Education Years=7', 'Education Years=8', 'Education Years=9', 'Education Years=10', 'Education Years=11', 'Education Years=12', 'Education Years=<6', 'Education Years=>12']\n"
     ]
    }
   ],
   "source": [
    "#splitting adult dataset \n",
    "#getting information for this dataset\n",
    "a_train, a_test, X_train_a, y_train_a, X_test_a, y_test_a, a_dataset_orig, privileged_groups, unprivileged_groups = splitdata (load_preproc_data_adult)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "Iop1fQk0CtvZ"
   },
   "source": [
    "### German"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "LgL63AM7TFAP",
    "outputId": "50437a0c-54d0-4cea-c052-34987eb1c065"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "training data size (700, 11)\n",
      "dataset feature names ['age', 'sex', 'credit_history=Delay', 'credit_history=None/Paid', 'credit_history=Other', 'savings=500+', 'savings=<500', 'savings=Unknown/None', 'employment=1-4 years', 'employment=4+ years', 'employment=Unemployed']\n",
      "The whole adult dataset shape is (1000, 11)\n",
      "The training adult dataset shape is (700, 11)\n",
      "The testing adult dataset shape is (300, 11)\n",
      "Favourable label is 1.0 and unfavourable label is 2.0\n",
      "Protected attribute names ['sex']\n",
      "Privileged protected attribute [array([1.])] and unpriviledged protected attribute [array([0.])]\n",
      "Dataset feature names ['age', 'sex', 'credit_history=Delay', 'credit_history=None/Paid', 'credit_history=Other', 'savings=500+', 'savings=<500', 'savings=Unknown/None', 'employment=1-4 years', 'employment=4+ years', 'employment=Unemployed']\n"
     ]
    }
   ],
   "source": [
    "#splitting german dataset \n",
    "#getting information for this dataset\n",
    "\n",
    "g_train, g_test, X_train_g, y_train_g, X_test_g, y_test_g, g_dataset_orig, privileged_groups, unprivileged_groups= splitdata (load_preproc_data_german)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "KNe0D2pwTdMi"
   },
   "source": [
    "### Explore"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "id": "RAzTj_U9vKrz"
   },
   "outputs": [],
   "source": [
    "#function to create a dataframe for further exploration\n",
    "\n",
    "def dataframe(data):\n",
    "  df = pd.DataFrame(np.hstack((data.features, data.labels)),\n",
    "                    columns = data.feature_names+data.label_names, \n",
    "                    index = data.instance_names)\n",
    "  df.loc[:,data.protected_attribute_names] = data.protected_attributes\n",
    "  return df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "4azdmeLCzE-H",
    "outputId": "52251eb1-0410-403d-869c-b2db7e2edf59"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(48842, 20)"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#what is the size of the adults dataset\n",
    "#adding the labels onto the df\n",
    "\n",
    "dfad = dataframe(a_dataset_orig)\n",
    "dfad['labels'] = a_dataset_orig.labels\n",
    "dfad.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "3I0fJpQkGlWm",
    "outputId": "9c98a703-515d-43f6-fa3e-e37c0143573f"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1000, 13)"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#what is the size of the adults dataset\n",
    "#adding the labels onto the df\n",
    "\n",
    "dfger = dataframe(g_dataset_orig)\n",
    "dfger['labels'] = g_dataset_orig.labels\n",
    "dfger.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "3HrTnYX7Cfge",
    "outputId": "f727efb4-421c-4189-a14c-da4bd349bd34"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "32650.0"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#this is therefore an unbalanced dataset\n",
    "\n",
    "sum(dfad['sex'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "cWIyI4J_Grku",
    "outputId": "b3c0d104-b8d8-49e9-bf41-861ef001fbc8"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "690.0"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#checking is the german dataset is also imbalanced with sex\n",
    "\n",
    "sum(dfger['sex'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 505
    },
    "id": "yYInurOwjS-z",
    "outputId": "d686d9f6-a57b-4e5f-e22b-bfe7c779bd72"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "\n",
       "  <div id=\"df-279b33ff-d87e-4b03-bdd8-f6e9f4527011\">\n",
       "    <div class=\"colab-df-container\">\n",
       "      <div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>race</th>\n",
       "      <th>sex</th>\n",
       "      <th>Age (decade)=10</th>\n",
       "      <th>Age (decade)=20</th>\n",
       "      <th>Age (decade)=30</th>\n",
       "      <th>Age (decade)=40</th>\n",
       "      <th>Age (decade)=50</th>\n",
       "      <th>Age (decade)=60</th>\n",
       "      <th>Age (decade)=&gt;=70</th>\n",
       "      <th>Education Years=6</th>\n",
       "      <th>Education Years=7</th>\n",
       "      <th>Education Years=8</th>\n",
       "      <th>Education Years=9</th>\n",
       "      <th>Education Years=10</th>\n",
       "      <th>Education Years=11</th>\n",
       "      <th>Education Years=12</th>\n",
       "      <th>Education Years=&lt;6</th>\n",
       "      <th>Education Years=&gt;12</th>\n",
       "      <th>Income Binary</th>\n",
       "      <th>labels</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>48837</th>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>48838</th>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>48839</th>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>48840</th>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>48841</th>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>48842 rows × 20 columns</p>\n",
       "</div>\n",
       "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-279b33ff-d87e-4b03-bdd8-f6e9f4527011')\"\n",
       "              title=\"Convert this dataframe to an interactive table.\"\n",
       "              style=\"display:none;\">\n",
       "        \n",
       "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
       "       width=\"24px\">\n",
       "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
       "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
       "  </svg>\n",
       "      </button>\n",
       "      \n",
       "  <style>\n",
       "    .colab-df-container {\n",
       "      display:flex;\n",
       "      flex-wrap:wrap;\n",
       "      gap: 12px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert {\n",
       "      background-color: #E8F0FE;\n",
       "      border: none;\n",
       "      border-radius: 50%;\n",
       "      cursor: pointer;\n",
       "      display: none;\n",
       "      fill: #1967D2;\n",
       "      height: 32px;\n",
       "      padding: 0 0 0 0;\n",
       "      width: 32px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert:hover {\n",
       "      background-color: #E2EBFA;\n",
       "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
       "      fill: #174EA6;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert {\n",
       "      background-color: #3B4455;\n",
       "      fill: #D2E3FC;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert:hover {\n",
       "      background-color: #434B5C;\n",
       "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
       "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
       "      fill: #FFFFFF;\n",
       "    }\n",
       "  </style>\n",
       "\n",
       "      <script>\n",
       "        const buttonEl =\n",
       "          document.querySelector('#df-279b33ff-d87e-4b03-bdd8-f6e9f4527011 button.colab-df-convert');\n",
       "        buttonEl.style.display =\n",
       "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
       "\n",
       "        async function convertToInteractive(key) {\n",
       "          const element = document.querySelector('#df-279b33ff-d87e-4b03-bdd8-f6e9f4527011');\n",
       "          const dataTable =\n",
       "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
       "                                                     [key], {});\n",
       "          if (!dataTable) return;\n",
       "\n",
       "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
       "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
       "            + ' to learn more about interactive tables.';\n",
       "          element.innerHTML = '';\n",
       "          dataTable['output_type'] = 'display_data';\n",
       "          await google.colab.output.renderOutput(dataTable, element);\n",
       "          const docLink = document.createElement('div');\n",
       "          docLink.innerHTML = docLinkHtml;\n",
       "          element.appendChild(docLink);\n",
       "        }\n",
       "      </script>\n",
       "    </div>\n",
       "  </div>\n",
       "  "
      ],
      "text/plain": [
       "       race  sex  Age (decade)=10  Age (decade)=20  Age (decade)=30  \\\n",
       "0       0.0  1.0              0.0              1.0              0.0   \n",
       "1       1.0  1.0              0.0              0.0              1.0   \n",
       "2       1.0  1.0              0.0              1.0              0.0   \n",
       "3       0.0  1.0              0.0              0.0              0.0   \n",
       "4       1.0  0.0              1.0              0.0              0.0   \n",
       "...     ...  ...              ...              ...              ...   \n",
       "48837   1.0  0.0              0.0              1.0              0.0   \n",
       "48838   1.0  1.0              0.0              0.0              0.0   \n",
       "48839   1.0  0.0              0.0              0.0              0.0   \n",
       "48840   1.0  1.0              0.0              1.0              0.0   \n",
       "48841   1.0  0.0              0.0              0.0              0.0   \n",
       "\n",
       "       Age (decade)=40  Age (decade)=50  Age (decade)=60  Age (decade)=>=70  \\\n",
       "0                  0.0              0.0              0.0                0.0   \n",
       "1                  0.0              0.0              0.0                0.0   \n",
       "2                  0.0              0.0              0.0                0.0   \n",
       "3                  1.0              0.0              0.0                0.0   \n",
       "4                  0.0              0.0              0.0                0.0   \n",
       "...                ...              ...              ...                ...   \n",
       "48837              0.0              0.0              0.0                0.0   \n",
       "48838              1.0              0.0              0.0                0.0   \n",
       "48839              0.0              1.0              0.0                0.0   \n",
       "48840              0.0              0.0              0.0                0.0   \n",
       "48841              0.0              1.0              0.0                0.0   \n",
       "\n",
       "       Education Years=6  Education Years=7  Education Years=8  \\\n",
       "0                    0.0                1.0                0.0   \n",
       "1                    0.0                0.0                0.0   \n",
       "2                    0.0                0.0                0.0   \n",
       "3                    0.0                0.0                0.0   \n",
       "4                    0.0                0.0                0.0   \n",
       "...                  ...                ...                ...   \n",
       "48837                0.0                0.0                0.0   \n",
       "48838                0.0                0.0                0.0   \n",
       "48839                0.0                0.0                0.0   \n",
       "48840                0.0                0.0                0.0   \n",
       "48841                0.0                0.0                0.0   \n",
       "\n",
       "       Education Years=9  Education Years=10  Education Years=11  \\\n",
       "0                    0.0                 0.0                 0.0   \n",
       "1                    1.0                 0.0                 0.0   \n",
       "2                    0.0                 0.0                 0.0   \n",
       "3                    0.0                 1.0                 0.0   \n",
       "4                    0.0                 1.0                 0.0   \n",
       "...                  ...                 ...                 ...   \n",
       "48837                0.0                 0.0                 0.0   \n",
       "48838                1.0                 0.0                 0.0   \n",
       "48839                1.0                 0.0                 0.0   \n",
       "48840                1.0                 0.0                 0.0   \n",
       "48841                1.0                 0.0                 0.0   \n",
       "\n",
       "       Education Years=12  Education Years=<6  Education Years=>12  \\\n",
       "0                     0.0                 0.0                  0.0   \n",
       "1                     0.0                 0.0                  0.0   \n",
       "2                     1.0                 0.0                  0.0   \n",
       "3                     0.0                 0.0                  0.0   \n",
       "4                     0.0                 0.0                  0.0   \n",
       "...                   ...                 ...                  ...   \n",
       "48837                 1.0                 0.0                  0.0   \n",
       "48838                 0.0                 0.0                  0.0   \n",
       "48839                 0.0                 0.0                  0.0   \n",
       "48840                 0.0                 0.0                  0.0   \n",
       "48841                 0.0                 0.0                  0.0   \n",
       "\n",
       "       Income Binary  labels  \n",
       "0                0.0     0.0  \n",
       "1                0.0     0.0  \n",
       "2                1.0     1.0  \n",
       "3                1.0     1.0  \n",
       "4                0.0     0.0  \n",
       "...              ...     ...  \n",
       "48837            0.0     0.0  \n",
       "48838            1.0     1.0  \n",
       "48839            0.0     0.0  \n",
       "48840            0.0     0.0  \n",
       "48841            1.0     1.0  \n",
       "\n",
       "[48842 rows x 20 columns]"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#printing the dataset to explore the other attributes \n",
    "dfad"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "jEm5Nx_ocnKh"
   },
   "source": [
    "## Picking a model "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "RpIYfQA0fXOh"
   },
   "source": [
    "This assignment will be studying the Logistic Regression model and observing the effects of regularisation on the accuracy-fairness tradeoff."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "wXHqV554EiPH"
   },
   "source": [
    "# **Task 1**\n",
    "\n",
    "The standard machine learning methods such as logistic regression, support vector machines, multi-layer perceptron use a trade-off hyperparameter (lambda, learning rate) to trade-off accuracy and generalisation (Lecture 5). The first task is to analyse whether or not better generalisation could correspond to fairer models.\n",
    "\n",
    "Specifically, for a given machine learning model of your choice, use training data to perform 5-fold cross validation. By varying the trade-off hyperparameter, select (1) the model with the highest accuracy, and (2) the model with the best fairness metric across 5 folds 1. After this model selection step, compute and report final accuracy and fairness metrics for both models (1) and (2) on the test set."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "3Wb7OzQ4HBs5"
   },
   "source": [
    "***\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "LwB8XY5LHHP6"
   },
   "source": [
    "## (1) Hyperparameter tuning using GridSearch\n",
    "\n",
    "In grid search, we preset a list of values for each hyperparameter. Then, we evaluate the model accuracy for every C value in this list."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "SC-tXliZZs2i"
   },
   "source": [
    "### Functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "id": "YRPe6WwqT98m"
   },
   "outputs": [],
   "source": [
    "#using grid-search for hyperparameter tuning\n",
    "\n",
    "#creating a function that inputs the x and y train values and changing c values\n",
    "def lr_gridsearch (x, y, cvalues):\n",
    "  #training the model\n",
    "  clf= GridSearchCV(LogisticRegression(solver = 'liblinear', random_state = 1), {'C':cvalues}, cv=5) #cv = 5 is using 5 fold cross validation\n",
    "  clf.fit(x, y)\n",
    "  clf.cv_results_ #gridsearch function that prints results\n",
    "\n",
    "  #displaying the results\n",
    "  #best params is a setting that gave the best results on the hold out data\n",
    "  #best score is the mean cross-validated score of the best_estimator \n",
    "  print(\"Tuned hyperparameter is at\",clf.best_params_, \"with accuracy :\",clf.best_score_) \n",
    "\n",
    "  #displaying the results\n",
    "  df = pd.DataFrame(clf.cv_results_)\n",
    "  display(df[['params','mean_test_score' ]])\n",
    "  \n",
    "  \n",
    "  xplt = np.array(cvalues)\n",
    "  plt.figure(figsize = (10,6))\n",
    "  plt.plot(xplt, df.mean_test_score)\n",
    "\n",
    "  #displaying the results in a graph\n",
    "  font1 = {'family':'serif','color':'blue','size':13}\n",
    "  font2 = {'family':'serif','color':'blue','size':11}\n",
    "  plt.title('Hyperparameter GridSearch Results', fontdict = font1)\n",
    "  plt.xlabel('C Values', fontdict = font2)\n",
    "  plt.ylabel('Accuracy', fontdict = font2)\n",
    "  plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "z94KqeguVucu"
   },
   "source": [
    "\n",
    "### Adult\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 642
    },
    "id": "scS5q4tQzIeJ",
    "outputId": "e1a41e9a-d1d8-42ef-cedc-ef272411b388"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tuned hyperparameter is at {'C': 0.05} with accuracy : 0.8043816906811315\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "\n",
       "  <div id=\"df-6c9cd461-b5d6-4754-875b-33a7b64837a4\">\n",
       "    <div class=\"colab-df-container\">\n",
       "      <div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>params</th>\n",
       "      <th>mean_test_score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>{'C': 0.005}</td>\n",
       "      <td>0.804323</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>{'C': 0.01}</td>\n",
       "      <td>0.804323</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>{'C': 0.05}</td>\n",
       "      <td>0.804382</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>{'C': 0.1}</td>\n",
       "      <td>0.804382</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>{'C': 0.5}</td>\n",
       "      <td>0.804382</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>{'C': 1}</td>\n",
       "      <td>0.804382</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>\n",
       "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-6c9cd461-b5d6-4754-875b-33a7b64837a4')\"\n",
       "              title=\"Convert this dataframe to an interactive table.\"\n",
       "              style=\"display:none;\">\n",
       "        \n",
       "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
       "       width=\"24px\">\n",
       "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
       "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
       "  </svg>\n",
       "      </button>\n",
       "      \n",
       "  <style>\n",
       "    .colab-df-container {\n",
       "      display:flex;\n",
       "      flex-wrap:wrap;\n",
       "      gap: 12px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert {\n",
       "      background-color: #E8F0FE;\n",
       "      border: none;\n",
       "      border-radius: 50%;\n",
       "      cursor: pointer;\n",
       "      display: none;\n",
       "      fill: #1967D2;\n",
       "      height: 32px;\n",
       "      padding: 0 0 0 0;\n",
       "      width: 32px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert:hover {\n",
       "      background-color: #E2EBFA;\n",
       "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
       "      fill: #174EA6;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert {\n",
       "      background-color: #3B4455;\n",
       "      fill: #D2E3FC;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert:hover {\n",
       "      background-color: #434B5C;\n",
       "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
       "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
       "      fill: #FFFFFF;\n",
       "    }\n",
       "  </style>\n",
       "\n",
       "      <script>\n",
       "        const buttonEl =\n",
       "          document.querySelector('#df-6c9cd461-b5d6-4754-875b-33a7b64837a4 button.colab-df-convert');\n",
       "        buttonEl.style.display =\n",
       "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
       "\n",
       "        async function convertToInteractive(key) {\n",
       "          const element = document.querySelector('#df-6c9cd461-b5d6-4754-875b-33a7b64837a4');\n",
       "          const dataTable =\n",
       "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
       "                                                     [key], {});\n",
       "          if (!dataTable) return;\n",
       "\n",
       "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
       "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
       "            + ' to learn more about interactive tables.';\n",
       "          element.innerHTML = '';\n",
       "          dataTable['output_type'] = 'display_data';\n",
       "          await google.colab.output.renderOutput(dataTable, element);\n",
       "          const docLink = document.createElement('div');\n",
       "          docLink.innerHTML = docLinkHtml;\n",
       "          element.appendChild(docLink);\n",
       "        }\n",
       "      </script>\n",
       "    </div>\n",
       "  </div>\n",
       "  "
      ],
      "text/plain": [
       "         params  mean_test_score\n",
       "0  {'C': 0.005}         0.804323\n",
       "1   {'C': 0.01}         0.804323\n",
       "2   {'C': 0.05}         0.804382\n",
       "3    {'C': 0.1}         0.804382\n",
       "4    {'C': 0.5}         0.804382\n",
       "5      {'C': 1}         0.804382"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#initially using a large range of c values with high distribution\n",
    "\n",
    "c = [0.005, 0.01, 0.05, 0.1, 0.5, 1]\n",
    "\n",
    "lr_gridsearch(X_train_a, y_train_a, c)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "Q2Ayw4Ypf7cz"
   },
   "source": [
    "Exploring the ***C*** values around 0.05 by creating a range of c values from 0.001 to 0.07"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "id": "RHSJ31-WUjyV",
    "outputId": "3490de0c-0fd1-44c6-b6cb-d850942feb03"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tuned hyperparameter is at {'C': 0.017} with accuracy : 0.8043816906811315\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "\n",
       "  <div id=\"df-d7b704d1-616e-46cf-900c-b992327b07dd\">\n",
       "    <div class=\"colab-df-container\">\n",
       "      <div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>params</th>\n",
       "      <th>mean_test_score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>{'C': 0.001}</td>\n",
       "      <td>0.803914</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>{'C': 0.003}</td>\n",
       "      <td>0.804060</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>{'C': 0.005}</td>\n",
       "      <td>0.804323</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>{'C': 0.007}</td>\n",
       "      <td>0.804323</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>{'C': 0.009000000000000001}</td>\n",
       "      <td>0.804323</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>{'C': 0.011}</td>\n",
       "      <td>0.804323</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>{'C': 0.013000000000000001}</td>\n",
       "      <td>0.804323</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>{'C': 0.015}</td>\n",
       "      <td>0.804323</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>{'C': 0.017}</td>\n",
       "      <td>0.804382</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>{'C': 0.019000000000000003}</td>\n",
       "      <td>0.804382</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>{'C': 0.021}</td>\n",
       "      <td>0.804382</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>{'C': 0.023}</td>\n",
       "      <td>0.804382</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>{'C': 0.025}</td>\n",
       "      <td>0.804382</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>{'C': 0.027000000000000003}</td>\n",
       "      <td>0.804382</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>{'C': 0.029}</td>\n",
       "      <td>0.804382</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>{'C': 0.031}</td>\n",
       "      <td>0.804382</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>{'C': 0.033}</td>\n",
       "      <td>0.804382</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>{'C': 0.035}</td>\n",
       "      <td>0.804382</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>{'C': 0.037000000000000005}</td>\n",
       "      <td>0.804382</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>{'C': 0.039}</td>\n",
       "      <td>0.804382</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>{'C': 0.041}</td>\n",
       "      <td>0.804382</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>{'C': 0.043000000000000003}</td>\n",
       "      <td>0.804382</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>{'C': 0.045}</td>\n",
       "      <td>0.804382</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>{'C': 0.047}</td>\n",
       "      <td>0.804382</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>{'C': 0.049}</td>\n",
       "      <td>0.804382</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>{'C': 0.051000000000000004}</td>\n",
       "      <td>0.804382</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>{'C': 0.053000000000000005}</td>\n",
       "      <td>0.804382</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>{'C': 0.055}</td>\n",
       "      <td>0.804382</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>{'C': 0.057}</td>\n",
       "      <td>0.804382</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>{'C': 0.059000000000000004}</td>\n",
       "      <td>0.804382</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>{'C': 0.061}</td>\n",
       "      <td>0.804382</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>31</th>\n",
       "      <td>{'C': 0.063}</td>\n",
       "      <td>0.804382</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32</th>\n",
       "      <td>{'C': 0.065}</td>\n",
       "      <td>0.804382</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33</th>\n",
       "      <td>{'C': 0.067}</td>\n",
       "      <td>0.804382</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>34</th>\n",
       "      <td>{'C': 0.069}</td>\n",
       "      <td>0.804382</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>\n",
       "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-d7b704d1-616e-46cf-900c-b992327b07dd')\"\n",
       "              title=\"Convert this dataframe to an interactive table.\"\n",
       "              style=\"display:none;\">\n",
       "        \n",
       "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
       "       width=\"24px\">\n",
       "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
       "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
       "  </svg>\n",
       "      </button>\n",
       "      \n",
       "  <style>\n",
       "    .colab-df-container {\n",
       "      display:flex;\n",
       "      flex-wrap:wrap;\n",
       "      gap: 12px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert {\n",
       "      background-color: #E8F0FE;\n",
       "      border: none;\n",
       "      border-radius: 50%;\n",
       "      cursor: pointer;\n",
       "      display: none;\n",
       "      fill: #1967D2;\n",
       "      height: 32px;\n",
       "      padding: 0 0 0 0;\n",
       "      width: 32px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert:hover {\n",
       "      background-color: #E2EBFA;\n",
       "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
       "      fill: #174EA6;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert {\n",
       "      background-color: #3B4455;\n",
       "      fill: #D2E3FC;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert:hover {\n",
       "      background-color: #434B5C;\n",
       "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
       "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
       "      fill: #FFFFFF;\n",
       "    }\n",
       "  </style>\n",
       "\n",
       "      <script>\n",
       "        const buttonEl =\n",
       "          document.querySelector('#df-d7b704d1-616e-46cf-900c-b992327b07dd button.colab-df-convert');\n",
       "        buttonEl.style.display =\n",
       "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
       "\n",
       "        async function convertToInteractive(key) {\n",
       "          const element = document.querySelector('#df-d7b704d1-616e-46cf-900c-b992327b07dd');\n",
       "          const dataTable =\n",
       "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
       "                                                     [key], {});\n",
       "          if (!dataTable) return;\n",
       "\n",
       "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
       "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
       "            + ' to learn more about interactive tables.';\n",
       "          element.innerHTML = '';\n",
       "          dataTable['output_type'] = 'display_data';\n",
       "          await google.colab.output.renderOutput(dataTable, element);\n",
       "          const docLink = document.createElement('div');\n",
       "          docLink.innerHTML = docLinkHtml;\n",
       "          element.appendChild(docLink);\n",
       "        }\n",
       "      </script>\n",
       "    </div>\n",
       "  </div>\n",
       "  "
      ],
      "text/plain": [
       "                         params  mean_test_score\n",
       "0                  {'C': 0.001}         0.803914\n",
       "1                  {'C': 0.003}         0.804060\n",
       "2                  {'C': 0.005}         0.804323\n",
       "3                  {'C': 0.007}         0.804323\n",
       "4   {'C': 0.009000000000000001}         0.804323\n",
       "5                  {'C': 0.011}         0.804323\n",
       "6   {'C': 0.013000000000000001}         0.804323\n",
       "7                  {'C': 0.015}         0.804323\n",
       "8                  {'C': 0.017}         0.804382\n",
       "9   {'C': 0.019000000000000003}         0.804382\n",
       "10                 {'C': 0.021}         0.804382\n",
       "11                 {'C': 0.023}         0.804382\n",
       "12                 {'C': 0.025}         0.804382\n",
       "13  {'C': 0.027000000000000003}         0.804382\n",
       "14                 {'C': 0.029}         0.804382\n",
       "15                 {'C': 0.031}         0.804382\n",
       "16                 {'C': 0.033}         0.804382\n",
       "17                 {'C': 0.035}         0.804382\n",
       "18  {'C': 0.037000000000000005}         0.804382\n",
       "19                 {'C': 0.039}         0.804382\n",
       "20                 {'C': 0.041}         0.804382\n",
       "21  {'C': 0.043000000000000003}         0.804382\n",
       "22                 {'C': 0.045}         0.804382\n",
       "23                 {'C': 0.047}         0.804382\n",
       "24                 {'C': 0.049}         0.804382\n",
       "25  {'C': 0.051000000000000004}         0.804382\n",
       "26  {'C': 0.053000000000000005}         0.804382\n",
       "27                 {'C': 0.055}         0.804382\n",
       "28                 {'C': 0.057}         0.804382\n",
       "29  {'C': 0.059000000000000004}         0.804382\n",
       "30                 {'C': 0.061}         0.804382\n",
       "31                 {'C': 0.063}         0.804382\n",
       "32                 {'C': 0.065}         0.804382\n",
       "33                 {'C': 0.067}         0.804382\n",
       "34                 {'C': 0.069}         0.804382"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#tuning into a more concentrated range of c\n",
    "\n",
    "a =  np.arange(0.001, 0.07, 0.002)\n",
    "\n",
    "lr_gridsearch(X_train_a, y_train_a, a)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "8Vru6yfg8dWF"
   },
   "source": [
    "\n",
    "**Adult Dataset:** The hyperparameter for the highest accuracy (**0.8043816906811315**) is at **C: 0.017**\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "e5x4sKYfVwGp"
   },
   "source": [
    "\n",
    "### German\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 642
    },
    "id": "p7sz1APQ0qTA",
    "outputId": "70d75416-cace-497c-b9a4-9d2c8023eb40"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tuned hyperparameter is at {'C': 0.01} with accuracy : 0.7185714285714286\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "\n",
       "  <div id=\"df-f5601f27-4903-45ed-9f04-b58d717a92f3\">\n",
       "    <div class=\"colab-df-container\">\n",
       "      <div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>params</th>\n",
       "      <th>mean_test_score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>{'C': 0.005}</td>\n",
       "      <td>0.715714</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>{'C': 0.01}</td>\n",
       "      <td>0.718571</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>{'C': 0.05}</td>\n",
       "      <td>0.715714</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>{'C': 0.1}</td>\n",
       "      <td>0.715714</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>{'C': 0.5}</td>\n",
       "      <td>0.715714</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>{'C': 1}</td>\n",
       "      <td>0.715714</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>\n",
       "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-f5601f27-4903-45ed-9f04-b58d717a92f3')\"\n",
       "              title=\"Convert this dataframe to an interactive table.\"\n",
       "              style=\"display:none;\">\n",
       "        \n",
       "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
       "       width=\"24px\">\n",
       "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
       "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
       "  </svg>\n",
       "      </button>\n",
       "      \n",
       "  <style>\n",
       "    .colab-df-container {\n",
       "      display:flex;\n",
       "      flex-wrap:wrap;\n",
       "      gap: 12px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert {\n",
       "      background-color: #E8F0FE;\n",
       "      border: none;\n",
       "      border-radius: 50%;\n",
       "      cursor: pointer;\n",
       "      display: none;\n",
       "      fill: #1967D2;\n",
       "      height: 32px;\n",
       "      padding: 0 0 0 0;\n",
       "      width: 32px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert:hover {\n",
       "      background-color: #E2EBFA;\n",
       "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
       "      fill: #174EA6;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert {\n",
       "      background-color: #3B4455;\n",
       "      fill: #D2E3FC;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert:hover {\n",
       "      background-color: #434B5C;\n",
       "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
       "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
       "      fill: #FFFFFF;\n",
       "    }\n",
       "  </style>\n",
       "\n",
       "      <script>\n",
       "        const buttonEl =\n",
       "          document.querySelector('#df-f5601f27-4903-45ed-9f04-b58d717a92f3 button.colab-df-convert');\n",
       "        buttonEl.style.display =\n",
       "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
       "\n",
       "        async function convertToInteractive(key) {\n",
       "          const element = document.querySelector('#df-f5601f27-4903-45ed-9f04-b58d717a92f3');\n",
       "          const dataTable =\n",
       "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
       "                                                     [key], {});\n",
       "          if (!dataTable) return;\n",
       "\n",
       "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
       "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
       "            + ' to learn more about interactive tables.';\n",
       "          element.innerHTML = '';\n",
       "          dataTable['output_type'] = 'display_data';\n",
       "          await google.colab.output.renderOutput(dataTable, element);\n",
       "          const docLink = document.createElement('div');\n",
       "          docLink.innerHTML = docLinkHtml;\n",
       "          element.appendChild(docLink);\n",
       "        }\n",
       "      </script>\n",
       "    </div>\n",
       "  </div>\n",
       "  "
      ],
      "text/plain": [
       "         params  mean_test_score\n",
       "0  {'C': 0.005}         0.715714\n",
       "1   {'C': 0.01}         0.718571\n",
       "2   {'C': 0.05}         0.715714\n",
       "3    {'C': 0.1}         0.715714\n",
       "4    {'C': 0.5}         0.715714\n",
       "5      {'C': 1}         0.715714"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#initially using a large range of c values with high distribution\n",
    "\n",
    "lr_gridsearch(X_train_g, y_train_g, c)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "MXvkli1r02Dn"
   },
   "source": [
    "Exploring the ***C*** values around 0.01 by creating a range of c values from 0.001 to 0.03"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "id": "fYKI-ZVHHHbZ",
    "outputId": "e1d6c7d0-5726-48bb-f2dd-6d5b8c4e28ba"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tuned hyperparameter is at {'C': 0.009000000000000001} with accuracy : 0.7185714285714286\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "\n",
       "  <div id=\"df-dc32ce56-153c-4321-aa07-51ef29ed67d7\">\n",
       "    <div class=\"colab-df-container\">\n",
       "      <div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>params</th>\n",
       "      <th>mean_test_score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>{'C': 0.001}</td>\n",
       "      <td>0.701429</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>{'C': 0.002}</td>\n",
       "      <td>0.708571</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>{'C': 0.003}</td>\n",
       "      <td>0.717143</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>{'C': 0.004}</td>\n",
       "      <td>0.714286</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>{'C': 0.005}</td>\n",
       "      <td>0.715714</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>{'C': 0.006}</td>\n",
       "      <td>0.717143</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>{'C': 0.007}</td>\n",
       "      <td>0.717143</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>{'C': 0.008}</td>\n",
       "      <td>0.717143</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>{'C': 0.009000000000000001}</td>\n",
       "      <td>0.718571</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>{'C': 0.010000000000000002}</td>\n",
       "      <td>0.718571</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>{'C': 0.011}</td>\n",
       "      <td>0.718571</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>{'C': 0.012}</td>\n",
       "      <td>0.714286</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>{'C': 0.013000000000000001}</td>\n",
       "      <td>0.715714</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>{'C': 0.014000000000000002}</td>\n",
       "      <td>0.715714</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>{'C': 0.015}</td>\n",
       "      <td>0.715714</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>{'C': 0.016}</td>\n",
       "      <td>0.717143</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>{'C': 0.017}</td>\n",
       "      <td>0.717143</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>{'C': 0.018000000000000002}</td>\n",
       "      <td>0.717143</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>{'C': 0.019000000000000003}</td>\n",
       "      <td>0.717143</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>{'C': 0.02}</td>\n",
       "      <td>0.715714</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>{'C': 0.021}</td>\n",
       "      <td>0.715714</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>{'C': 0.022000000000000002}</td>\n",
       "      <td>0.715714</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>{'C': 0.023}</td>\n",
       "      <td>0.715714</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>{'C': 0.024}</td>\n",
       "      <td>0.715714</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>{'C': 0.025}</td>\n",
       "      <td>0.715714</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>{'C': 0.026000000000000002}</td>\n",
       "      <td>0.715714</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>{'C': 0.027000000000000003}</td>\n",
       "      <td>0.715714</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>{'C': 0.028}</td>\n",
       "      <td>0.715714</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>{'C': 0.029}</td>\n",
       "      <td>0.715714</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>\n",
       "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-dc32ce56-153c-4321-aa07-51ef29ed67d7')\"\n",
       "              title=\"Convert this dataframe to an interactive table.\"\n",
       "              style=\"display:none;\">\n",
       "        \n",
       "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
       "       width=\"24px\">\n",
       "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
       "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
       "  </svg>\n",
       "      </button>\n",
       "      \n",
       "  <style>\n",
       "    .colab-df-container {\n",
       "      display:flex;\n",
       "      flex-wrap:wrap;\n",
       "      gap: 12px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert {\n",
       "      background-color: #E8F0FE;\n",
       "      border: none;\n",
       "      border-radius: 50%;\n",
       "      cursor: pointer;\n",
       "      display: none;\n",
       "      fill: #1967D2;\n",
       "      height: 32px;\n",
       "      padding: 0 0 0 0;\n",
       "      width: 32px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert:hover {\n",
       "      background-color: #E2EBFA;\n",
       "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
       "      fill: #174EA6;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert {\n",
       "      background-color: #3B4455;\n",
       "      fill: #D2E3FC;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert:hover {\n",
       "      background-color: #434B5C;\n",
       "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
       "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
       "      fill: #FFFFFF;\n",
       "    }\n",
       "  </style>\n",
       "\n",
       "      <script>\n",
       "        const buttonEl =\n",
       "          document.querySelector('#df-dc32ce56-153c-4321-aa07-51ef29ed67d7 button.colab-df-convert');\n",
       "        buttonEl.style.display =\n",
       "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
       "\n",
       "        async function convertToInteractive(key) {\n",
       "          const element = document.querySelector('#df-dc32ce56-153c-4321-aa07-51ef29ed67d7');\n",
       "          const dataTable =\n",
       "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
       "                                                     [key], {});\n",
       "          if (!dataTable) return;\n",
       "\n",
       "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
       "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
       "            + ' to learn more about interactive tables.';\n",
       "          element.innerHTML = '';\n",
       "          dataTable['output_type'] = 'display_data';\n",
       "          await google.colab.output.renderOutput(dataTable, element);\n",
       "          const docLink = document.createElement('div');\n",
       "          docLink.innerHTML = docLinkHtml;\n",
       "          element.appendChild(docLink);\n",
       "        }\n",
       "      </script>\n",
       "    </div>\n",
       "  </div>\n",
       "  "
      ],
      "text/plain": [
       "                         params  mean_test_score\n",
       "0                  {'C': 0.001}         0.701429\n",
       "1                  {'C': 0.002}         0.708571\n",
       "2                  {'C': 0.003}         0.717143\n",
       "3                  {'C': 0.004}         0.714286\n",
       "4                  {'C': 0.005}         0.715714\n",
       "5                  {'C': 0.006}         0.717143\n",
       "6                  {'C': 0.007}         0.717143\n",
       "7                  {'C': 0.008}         0.717143\n",
       "8   {'C': 0.009000000000000001}         0.718571\n",
       "9   {'C': 0.010000000000000002}         0.718571\n",
       "10                 {'C': 0.011}         0.718571\n",
       "11                 {'C': 0.012}         0.714286\n",
       "12  {'C': 0.013000000000000001}         0.715714\n",
       "13  {'C': 0.014000000000000002}         0.715714\n",
       "14                 {'C': 0.015}         0.715714\n",
       "15                 {'C': 0.016}         0.717143\n",
       "16                 {'C': 0.017}         0.717143\n",
       "17  {'C': 0.018000000000000002}         0.717143\n",
       "18  {'C': 0.019000000000000003}         0.717143\n",
       "19                  {'C': 0.02}         0.715714\n",
       "20                 {'C': 0.021}         0.715714\n",
       "21  {'C': 0.022000000000000002}         0.715714\n",
       "22                 {'C': 0.023}         0.715714\n",
       "23                 {'C': 0.024}         0.715714\n",
       "24                 {'C': 0.025}         0.715714\n",
       "25  {'C': 0.026000000000000002}         0.715714\n",
       "26  {'C': 0.027000000000000003}         0.715714\n",
       "27                 {'C': 0.028}         0.715714\n",
       "28                 {'C': 0.029}         0.715714"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "b =  np.arange(0.001, 0.03, 0.001)\n",
    "\n",
    "lr_gridsearch(X_train_g, y_train_g, b)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "UUTaacjV8S3N"
   },
   "source": [
    "**German Dataset:** The hyperparameter for the highest accuracy (**0.7185714285714286**) is at **C: 0.009**\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "sIDdV2n-9NHd"
   },
   "source": [
    "## (2) Equality of Opportunity Fairness Metrics\n",
    "\n",
    "Adding Equality of Opportunity into our function, combing accuracy and Equality of Opportunity\n",
    "\n",
    "The fairness metric we use is the Equality of Opportunity \n",
    "- Equality of Opportunity: True Negative Rate (TNR) or True Positive Rate (TPR) should be equal for all groups\n",
    "\n",
    "\n",
    "Ideally we want to satisy all fairness definitions but this is only possible under trivial circumstances."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "HVszlrHgZmMa"
   },
   "source": [
    "### Functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "id": "Wfa180lfVkKg"
   },
   "outputs": [],
   "source": [
    "#function to evaluate the accuracy and the fairness of each hyperparameter and report the best\n",
    "\n",
    "def cv_accuracy_fairness(X_train, y_train, train, cvals):\n",
    "  c_values = cvals\n",
    "  #empty lists\n",
    "  acc_scores = []\n",
    "  equ_opp = []\n",
    "  avg_acc_scores = []\n",
    "\n",
    "  #a loop to evaluate each hyperparameter under 5-fold cross validation\n",
    "  for i in c_values:\n",
    "    #training the model\n",
    "    clf = LogisticRegression(solver='liblinear', C=i, random_state=1)\n",
    "    #using cross val score to find the accuracy, can be used instead of gridsearch\n",
    "    score = cross_val_score(clf, X_train, y_train, cv = 5)\n",
    "    #using cross val predict to generate cross-validated estimates for each input data point \n",
    "    y_fair = cross_val_predict(clf, X_train, y_train, cv = 5)\n",
    "    test_pred = train.copy()\n",
    "    test_pred.labels = y_fair\n",
    "    metric = ClassificationMetric(train, test_pred, unprivileged_groups=unprivileged_groups, privileged_groups=privileged_groups)\n",
    "    equ_opp.append(metric.equal_opportunity_difference())\n",
    "\n",
    "    #print(equ_opp)\n",
    "    avg_acc = np.mean(score)\n",
    "    avg_acc_scores.append(np.mean(score))\n",
    "  \n",
    "  #printing the best of the equal opportunity scores\n",
    "  equal_opp_best = min(equ_opp, key= abs)\n",
    "  #for the best c value\n",
    "  best_c = equ_opp.index(equal_opp_best)\n",
    "  #best average score\n",
    "  #this should be the same as the grid search\n",
    "  best_acc_c = avg_acc_scores.index(max(avg_acc_scores))\n",
    "\n",
    "  #printing off the results and corresponding c value\n",
    "  print('Best Equality of Opportunity score is at C: ', c_values[best_c], 'with Equality of Opportunity score of', equal_opp_best)\n",
    "  print('Best accuracy score is at C: ', c_values[best_acc_c], 'with accuracy score of', max(avg_acc_scores))\n",
    "\n",
    "  #creating a df to show the values\n",
    "  df = pd.DataFrame({'C': c_values, \n",
    "                  'Accuracy' :avg_acc_scores,\n",
    "                  'Equality of Opportunity': equ_opp})\n",
    "  \n",
    "  #plotting the results \n",
    "  xplt = np.array(cvals)\n",
    "  plt.figure(figsize = (10,6))\n",
    "  plt.plot(xplt, df['Equality of Opportunity'], color = 'green')\n",
    "  font1 = {'family':'serif','color':'green','size':13}\n",
    "  font2 = {'family':'serif','color':'green','size':11}\n",
    "  plt.title('Equality of Opportunity Hyperparameter Tuning', fontdict = font1)\n",
    "  plt.xlabel('C Value', fontdict = font2)\n",
    "  plt.ylabel('Equality of Opportunity', fontdict = font2)\n",
    "  plt.show()\n",
    "  return df\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "ItOeCeTuNqlb"
   },
   "source": [
    "### Adult\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 439
    },
    "id": "fJgOmnr4c6Co",
    "outputId": "0d57cbbb-2e4d-48d7-910c-086055951493"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best Equality of Opportunity score is at C:  0.001 with Equality of Opportunity score of -0.4464800923254472\n",
      "Best accuracy score is at C:  0.017 with accuracy score of 0.8043816906811315\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "dfa = cv_accuracy_fairness(X_train_a, y_train_a, a_train, a)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "id": "sIImxMLKWMl5",
    "outputId": "38cfc996-91d5-41dd-812a-877f89c9f8e7"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style type=\"text/css\">\n",
       "#T_74536_row0_col0, #T_74536_row0_col1, #T_74536_row0_col2 {\n",
       "  background-color: lightgreen;\n",
       "}\n",
       "</style>\n",
       "<table id=\"T_74536_\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th class=\"blank level0\" >&nbsp;</th>\n",
       "      <th class=\"col_heading level0 col0\" >C</th>\n",
       "      <th class=\"col_heading level0 col1\" >Accuracy</th>\n",
       "      <th class=\"col_heading level0 col2\" >Equality of Opportunity</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row0\" class=\"row_heading level0 row0\" >0</th>\n",
       "      <td id=\"T_74536_row0_col0\" class=\"data row0 col0\" >0.001000</td>\n",
       "      <td id=\"T_74536_row0_col1\" class=\"data row0 col1\" >0.803914</td>\n",
       "      <td id=\"T_74536_row0_col2\" class=\"data row0 col2\" >-0.446480</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row1\" class=\"row_heading level0 row1\" >1</th>\n",
       "      <td id=\"T_74536_row1_col0\" class=\"data row1 col0\" >0.003000</td>\n",
       "      <td id=\"T_74536_row1_col1\" class=\"data row1 col1\" >0.804060</td>\n",
       "      <td id=\"T_74536_row1_col2\" class=\"data row1 col2\" >-0.453693</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row2\" class=\"row_heading level0 row2\" >2</th>\n",
       "      <td id=\"T_74536_row2_col0\" class=\"data row2 col0\" >0.005000</td>\n",
       "      <td id=\"T_74536_row2_col1\" class=\"data row2 col1\" >0.804323</td>\n",
       "      <td id=\"T_74536_row2_col2\" class=\"data row2 col2\" >-0.457155</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row3\" class=\"row_heading level0 row3\" >3</th>\n",
       "      <td id=\"T_74536_row3_col0\" class=\"data row3 col0\" >0.007000</td>\n",
       "      <td id=\"T_74536_row3_col1\" class=\"data row3 col1\" >0.804323</td>\n",
       "      <td id=\"T_74536_row3_col2\" class=\"data row3 col2\" >-0.457155</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row4\" class=\"row_heading level0 row4\" >4</th>\n",
       "      <td id=\"T_74536_row4_col0\" class=\"data row4 col0\" >0.009000</td>\n",
       "      <td id=\"T_74536_row4_col1\" class=\"data row4 col1\" >0.804323</td>\n",
       "      <td id=\"T_74536_row4_col2\" class=\"data row4 col2\" >-0.457155</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row5\" class=\"row_heading level0 row5\" >5</th>\n",
       "      <td id=\"T_74536_row5_col0\" class=\"data row5 col0\" >0.011000</td>\n",
       "      <td id=\"T_74536_row5_col1\" class=\"data row5 col1\" >0.804323</td>\n",
       "      <td id=\"T_74536_row5_col2\" class=\"data row5 col2\" >-0.457155</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row6\" class=\"row_heading level0 row6\" >6</th>\n",
       "      <td id=\"T_74536_row6_col0\" class=\"data row6 col0\" >0.013000</td>\n",
       "      <td id=\"T_74536_row6_col1\" class=\"data row6 col1\" >0.804323</td>\n",
       "      <td id=\"T_74536_row6_col2\" class=\"data row6 col2\" >-0.457155</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row7\" class=\"row_heading level0 row7\" >7</th>\n",
       "      <td id=\"T_74536_row7_col0\" class=\"data row7 col0\" >0.015000</td>\n",
       "      <td id=\"T_74536_row7_col1\" class=\"data row7 col1\" >0.804323</td>\n",
       "      <td id=\"T_74536_row7_col2\" class=\"data row7 col2\" >-0.457155</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row8\" class=\"row_heading level0 row8\" >8</th>\n",
       "      <td id=\"T_74536_row8_col0\" class=\"data row8 col0\" >0.017000</td>\n",
       "      <td id=\"T_74536_row8_col1\" class=\"data row8 col1\" >0.804382</td>\n",
       "      <td id=\"T_74536_row8_col2\" class=\"data row8 col2\" >-0.457588</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row9\" class=\"row_heading level0 row9\" >9</th>\n",
       "      <td id=\"T_74536_row9_col0\" class=\"data row9 col0\" >0.019000</td>\n",
       "      <td id=\"T_74536_row9_col1\" class=\"data row9 col1\" >0.804382</td>\n",
       "      <td id=\"T_74536_row9_col2\" class=\"data row9 col2\" >-0.457588</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row10\" class=\"row_heading level0 row10\" >10</th>\n",
       "      <td id=\"T_74536_row10_col0\" class=\"data row10 col0\" >0.021000</td>\n",
       "      <td id=\"T_74536_row10_col1\" class=\"data row10 col1\" >0.804382</td>\n",
       "      <td id=\"T_74536_row10_col2\" class=\"data row10 col2\" >-0.457588</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row11\" class=\"row_heading level0 row11\" >11</th>\n",
       "      <td id=\"T_74536_row11_col0\" class=\"data row11 col0\" >0.023000</td>\n",
       "      <td id=\"T_74536_row11_col1\" class=\"data row11 col1\" >0.804382</td>\n",
       "      <td id=\"T_74536_row11_col2\" class=\"data row11 col2\" >-0.457588</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row12\" class=\"row_heading level0 row12\" >12</th>\n",
       "      <td id=\"T_74536_row12_col0\" class=\"data row12 col0\" >0.025000</td>\n",
       "      <td id=\"T_74536_row12_col1\" class=\"data row12 col1\" >0.804382</td>\n",
       "      <td id=\"T_74536_row12_col2\" class=\"data row12 col2\" >-0.457588</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row13\" class=\"row_heading level0 row13\" >13</th>\n",
       "      <td id=\"T_74536_row13_col0\" class=\"data row13 col0\" >0.027000</td>\n",
       "      <td id=\"T_74536_row13_col1\" class=\"data row13 col1\" >0.804382</td>\n",
       "      <td id=\"T_74536_row13_col2\" class=\"data row13 col2\" >-0.457588</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row14\" class=\"row_heading level0 row14\" >14</th>\n",
       "      <td id=\"T_74536_row14_col0\" class=\"data row14 col0\" >0.029000</td>\n",
       "      <td id=\"T_74536_row14_col1\" class=\"data row14 col1\" >0.804382</td>\n",
       "      <td id=\"T_74536_row14_col2\" class=\"data row14 col2\" >-0.457588</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row15\" class=\"row_heading level0 row15\" >15</th>\n",
       "      <td id=\"T_74536_row15_col0\" class=\"data row15 col0\" >0.031000</td>\n",
       "      <td id=\"T_74536_row15_col1\" class=\"data row15 col1\" >0.804382</td>\n",
       "      <td id=\"T_74536_row15_col2\" class=\"data row15 col2\" >-0.457588</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row16\" class=\"row_heading level0 row16\" >16</th>\n",
       "      <td id=\"T_74536_row16_col0\" class=\"data row16 col0\" >0.033000</td>\n",
       "      <td id=\"T_74536_row16_col1\" class=\"data row16 col1\" >0.804382</td>\n",
       "      <td id=\"T_74536_row16_col2\" class=\"data row16 col2\" >-0.457588</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row17\" class=\"row_heading level0 row17\" >17</th>\n",
       "      <td id=\"T_74536_row17_col0\" class=\"data row17 col0\" >0.035000</td>\n",
       "      <td id=\"T_74536_row17_col1\" class=\"data row17 col1\" >0.804382</td>\n",
       "      <td id=\"T_74536_row17_col2\" class=\"data row17 col2\" >-0.457588</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row18\" class=\"row_heading level0 row18\" >18</th>\n",
       "      <td id=\"T_74536_row18_col0\" class=\"data row18 col0\" >0.037000</td>\n",
       "      <td id=\"T_74536_row18_col1\" class=\"data row18 col1\" >0.804382</td>\n",
       "      <td id=\"T_74536_row18_col2\" class=\"data row18 col2\" >-0.457588</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row19\" class=\"row_heading level0 row19\" >19</th>\n",
       "      <td id=\"T_74536_row19_col0\" class=\"data row19 col0\" >0.039000</td>\n",
       "      <td id=\"T_74536_row19_col1\" class=\"data row19 col1\" >0.804382</td>\n",
       "      <td id=\"T_74536_row19_col2\" class=\"data row19 col2\" >-0.457588</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row20\" class=\"row_heading level0 row20\" >20</th>\n",
       "      <td id=\"T_74536_row20_col0\" class=\"data row20 col0\" >0.041000</td>\n",
       "      <td id=\"T_74536_row20_col1\" class=\"data row20 col1\" >0.804382</td>\n",
       "      <td id=\"T_74536_row20_col2\" class=\"data row20 col2\" >-0.457588</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row21\" class=\"row_heading level0 row21\" >21</th>\n",
       "      <td id=\"T_74536_row21_col0\" class=\"data row21 col0\" >0.043000</td>\n",
       "      <td id=\"T_74536_row21_col1\" class=\"data row21 col1\" >0.804382</td>\n",
       "      <td id=\"T_74536_row21_col2\" class=\"data row21 col2\" >-0.457588</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row22\" class=\"row_heading level0 row22\" >22</th>\n",
       "      <td id=\"T_74536_row22_col0\" class=\"data row22 col0\" >0.045000</td>\n",
       "      <td id=\"T_74536_row22_col1\" class=\"data row22 col1\" >0.804382</td>\n",
       "      <td id=\"T_74536_row22_col2\" class=\"data row22 col2\" >-0.457588</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row23\" class=\"row_heading level0 row23\" >23</th>\n",
       "      <td id=\"T_74536_row23_col0\" class=\"data row23 col0\" >0.047000</td>\n",
       "      <td id=\"T_74536_row23_col1\" class=\"data row23 col1\" >0.804382</td>\n",
       "      <td id=\"T_74536_row23_col2\" class=\"data row23 col2\" >-0.457588</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row24\" class=\"row_heading level0 row24\" >24</th>\n",
       "      <td id=\"T_74536_row24_col0\" class=\"data row24 col0\" >0.049000</td>\n",
       "      <td id=\"T_74536_row24_col1\" class=\"data row24 col1\" >0.804382</td>\n",
       "      <td id=\"T_74536_row24_col2\" class=\"data row24 col2\" >-0.457588</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row25\" class=\"row_heading level0 row25\" >25</th>\n",
       "      <td id=\"T_74536_row25_col0\" class=\"data row25 col0\" >0.051000</td>\n",
       "      <td id=\"T_74536_row25_col1\" class=\"data row25 col1\" >0.804382</td>\n",
       "      <td id=\"T_74536_row25_col2\" class=\"data row25 col2\" >-0.457588</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row26\" class=\"row_heading level0 row26\" >26</th>\n",
       "      <td id=\"T_74536_row26_col0\" class=\"data row26 col0\" >0.053000</td>\n",
       "      <td id=\"T_74536_row26_col1\" class=\"data row26 col1\" >0.804382</td>\n",
       "      <td id=\"T_74536_row26_col2\" class=\"data row26 col2\" >-0.457588</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row27\" class=\"row_heading level0 row27\" >27</th>\n",
       "      <td id=\"T_74536_row27_col0\" class=\"data row27 col0\" >0.055000</td>\n",
       "      <td id=\"T_74536_row27_col1\" class=\"data row27 col1\" >0.804382</td>\n",
       "      <td id=\"T_74536_row27_col2\" class=\"data row27 col2\" >-0.457588</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row28\" class=\"row_heading level0 row28\" >28</th>\n",
       "      <td id=\"T_74536_row28_col0\" class=\"data row28 col0\" >0.057000</td>\n",
       "      <td id=\"T_74536_row28_col1\" class=\"data row28 col1\" >0.804382</td>\n",
       "      <td id=\"T_74536_row28_col2\" class=\"data row28 col2\" >-0.457588</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row29\" class=\"row_heading level0 row29\" >29</th>\n",
       "      <td id=\"T_74536_row29_col0\" class=\"data row29 col0\" >0.059000</td>\n",
       "      <td id=\"T_74536_row29_col1\" class=\"data row29 col1\" >0.804382</td>\n",
       "      <td id=\"T_74536_row29_col2\" class=\"data row29 col2\" >-0.457588</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row30\" class=\"row_heading level0 row30\" >30</th>\n",
       "      <td id=\"T_74536_row30_col0\" class=\"data row30 col0\" >0.061000</td>\n",
       "      <td id=\"T_74536_row30_col1\" class=\"data row30 col1\" >0.804382</td>\n",
       "      <td id=\"T_74536_row30_col2\" class=\"data row30 col2\" >-0.457588</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row31\" class=\"row_heading level0 row31\" >31</th>\n",
       "      <td id=\"T_74536_row31_col0\" class=\"data row31 col0\" >0.063000</td>\n",
       "      <td id=\"T_74536_row31_col1\" class=\"data row31 col1\" >0.804382</td>\n",
       "      <td id=\"T_74536_row31_col2\" class=\"data row31 col2\" >-0.457588</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row32\" class=\"row_heading level0 row32\" >32</th>\n",
       "      <td id=\"T_74536_row32_col0\" class=\"data row32 col0\" >0.065000</td>\n",
       "      <td id=\"T_74536_row32_col1\" class=\"data row32 col1\" >0.804382</td>\n",
       "      <td id=\"T_74536_row32_col2\" class=\"data row32 col2\" >-0.457588</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row33\" class=\"row_heading level0 row33\" >33</th>\n",
       "      <td id=\"T_74536_row33_col0\" class=\"data row33 col0\" >0.067000</td>\n",
       "      <td id=\"T_74536_row33_col1\" class=\"data row33 col1\" >0.804382</td>\n",
       "      <td id=\"T_74536_row33_col2\" class=\"data row33 col2\" >-0.457588</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_74536_level0_row34\" class=\"row_heading level0 row34\" >34</th>\n",
       "      <td id=\"T_74536_row34_col0\" class=\"data row34 col0\" >0.069000</td>\n",
       "      <td id=\"T_74536_row34_col1\" class=\"data row34 col1\" >0.804382</td>\n",
       "      <td id=\"T_74536_row34_col2\" class=\"data row34 col2\" >-0.457588</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n"
      ],
      "text/plain": [
       "<pandas.io.formats.style.Styler at 0x7f32c6e383d0>"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#highlighting the best c value\n",
    "\n",
    "def yellow(s):\n",
    "  if s['C'] == 0.001:\n",
    "    return ['background-color: lightgreen']*3\n",
    "\n",
    "dfa.style.apply(yellow, axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "quxTwz1_9Hr8"
   },
   "source": [
    "Refining the C values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 439
    },
    "id": "F0fAjusx7sm8",
    "outputId": "2e6962bb-4468-4a0e-e088-52e8a58ee33d"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best Equality of Opportunity score is at C:  0.001 with Equality of Opportunity score of -0.4464800923254472\n",
      "Best accuracy score is at C:  0.004 with accuracy score of 0.8043231940446883\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#narrowing in and expanding the c values\n",
    "\n",
    "dfa2 = cv_accuracy_fairness(X_train_a, y_train_a, a_train, np.arange(0.001, 0.005, 0.0005))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "c-aqse9G-LvY"
   },
   "source": [
    "**Adult Dataset:** The hyperparameter for the best *Equality of Opportunity* is at **C: 0.001** with a score of **-0.44648009**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "QPzc6swdVgDU"
   },
   "source": [
    "### German\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 439
    },
    "id": "_FU9ql3_WbQu",
    "outputId": "5ab90e1e-e903-492f-839e-86fea9a799cd"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best Equality of Opportunity score is at C:  0.051000000000000004 with Equality of Opportunity score of -0.050097847358121395\n",
      "Best accuracy score is at C:  0.009000000000000001 with accuracy score of 0.7185714285714286\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#for german\n",
    "\n",
    "dfg = cv_accuracy_fairness(X_train_g, y_train_g, g_train, a)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "GzqpNTUt-_l1"
   },
   "source": [
    "**German Dataset:** The hyperparameter for the best *Equality of Opportunity* is at **C: 0.051** with a score of **-0.050097847358121395**\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "id": "f1sBhjCmXAun",
    "outputId": "287d0140-e04c-43b4-9f65-72830b1f878a"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style type=\"text/css\">\n",
       "#T_78eee_row0_col0, #T_78eee_row0_col1, #T_78eee_row0_col2, #T_78eee_row1_col0, #T_78eee_row1_col1, #T_78eee_row1_col2, #T_78eee_row2_col0, #T_78eee_row2_col1, #T_78eee_row2_col2, #T_78eee_row3_col0, #T_78eee_row3_col1, #T_78eee_row3_col2, #T_78eee_row4_col0, #T_78eee_row4_col1, #T_78eee_row4_col2, #T_78eee_row5_col0, #T_78eee_row5_col1, #T_78eee_row5_col2, #T_78eee_row6_col0, #T_78eee_row6_col1, #T_78eee_row6_col2, #T_78eee_row7_col0, #T_78eee_row7_col1, #T_78eee_row7_col2, #T_78eee_row8_col0, #T_78eee_row8_col1, #T_78eee_row8_col2, #T_78eee_row9_col0, #T_78eee_row9_col1, #T_78eee_row9_col2, #T_78eee_row10_col0, #T_78eee_row10_col1, #T_78eee_row10_col2, #T_78eee_row11_col0, #T_78eee_row11_col1, #T_78eee_row11_col2, #T_78eee_row12_col0, #T_78eee_row12_col1, #T_78eee_row12_col2, #T_78eee_row13_col0, #T_78eee_row13_col1, #T_78eee_row13_col2, #T_78eee_row14_col0, #T_78eee_row14_col1, #T_78eee_row14_col2, #T_78eee_row15_col0, #T_78eee_row15_col1, #T_78eee_row15_col2, #T_78eee_row16_col0, #T_78eee_row16_col1, #T_78eee_row16_col2, #T_78eee_row17_col0, #T_78eee_row17_col1, #T_78eee_row17_col2, #T_78eee_row18_col0, #T_78eee_row18_col1, #T_78eee_row18_col2, #T_78eee_row19_col0, #T_78eee_row19_col1, #T_78eee_row19_col2, #T_78eee_row20_col0, #T_78eee_row20_col1, #T_78eee_row20_col2, #T_78eee_row21_col0, #T_78eee_row21_col1, #T_78eee_row21_col2, #T_78eee_row22_col0, #T_78eee_row22_col1, #T_78eee_row22_col2, #T_78eee_row23_col0, #T_78eee_row23_col1, #T_78eee_row23_col2, #T_78eee_row24_col0, #T_78eee_row24_col1, #T_78eee_row24_col2, #T_78eee_row26_col0, #T_78eee_row26_col1, #T_78eee_row26_col2, #T_78eee_row27_col0, #T_78eee_row27_col1, #T_78eee_row27_col2, #T_78eee_row28_col0, #T_78eee_row28_col1, #T_78eee_row28_col2, #T_78eee_row29_col0, #T_78eee_row29_col1, #T_78eee_row29_col2, #T_78eee_row30_col0, #T_78eee_row30_col1, #T_78eee_row30_col2, #T_78eee_row31_col0, #T_78eee_row31_col1, #T_78eee_row31_col2, #T_78eee_row32_col0, #T_78eee_row32_col1, #T_78eee_row32_col2, #T_78eee_row33_col0, #T_78eee_row33_col1, #T_78eee_row33_col2, #T_78eee_row34_col0, #T_78eee_row34_col1, #T_78eee_row34_col2 {\n",
       "  background-color: white;\n",
       "}\n",
       "#T_78eee_row25_col0, #T_78eee_row25_col1, #T_78eee_row25_col2 {\n",
       "  background-color: lightgreen;\n",
       "}\n",
       "</style>\n",
       "<table id=\"T_78eee_\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th class=\"blank level0\" >&nbsp;</th>\n",
       "      <th class=\"col_heading level0 col0\" >C</th>\n",
       "      <th class=\"col_heading level0 col1\" >Accuracy</th>\n",
       "      <th class=\"col_heading level0 col2\" >Equality of Opportunity</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row0\" class=\"row_heading level0 row0\" >0</th>\n",
       "      <td id=\"T_78eee_row0_col0\" class=\"data row0 col0\" >0.001000</td>\n",
       "      <td id=\"T_78eee_row0_col1\" class=\"data row0 col1\" >0.701429</td>\n",
       "      <td id=\"T_78eee_row0_col2\" class=\"data row0 col2\" >-0.191977</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row1\" class=\"row_heading level0 row1\" >1</th>\n",
       "      <td id=\"T_78eee_row1_col0\" class=\"data row1 col0\" >0.003000</td>\n",
       "      <td id=\"T_78eee_row1_col1\" class=\"data row1 col1\" >0.717143</td>\n",
       "      <td id=\"T_78eee_row1_col2\" class=\"data row1 col2\" >-0.053523</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row2\" class=\"row_heading level0 row2\" >2</th>\n",
       "      <td id=\"T_78eee_row2_col0\" class=\"data row2 col0\" >0.005000</td>\n",
       "      <td id=\"T_78eee_row2_col1\" class=\"data row2 col1\" >0.715714</td>\n",
       "      <td id=\"T_78eee_row2_col2\" class=\"data row2 col2\" >-0.064481</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row3\" class=\"row_heading level0 row3\" >3</th>\n",
       "      <td id=\"T_78eee_row3_col0\" class=\"data row3 col0\" >0.007000</td>\n",
       "      <td id=\"T_78eee_row3_col1\" class=\"data row3 col1\" >0.717143</td>\n",
       "      <td id=\"T_78eee_row3_col2\" class=\"data row3 col2\" >-0.075440</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row4\" class=\"row_heading level0 row4\" >4</th>\n",
       "      <td id=\"T_78eee_row4_col0\" class=\"data row4 col0\" >0.009000</td>\n",
       "      <td id=\"T_78eee_row4_col1\" class=\"data row4 col1\" >0.718571</td>\n",
       "      <td id=\"T_78eee_row4_col2\" class=\"data row4 col2\" >-0.072114</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row5\" class=\"row_heading level0 row5\" >5</th>\n",
       "      <td id=\"T_78eee_row5_col0\" class=\"data row5 col0\" >0.011000</td>\n",
       "      <td id=\"T_78eee_row5_col1\" class=\"data row5 col1\" >0.718571</td>\n",
       "      <td id=\"T_78eee_row5_col2\" class=\"data row5 col2\" >-0.072114</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row6\" class=\"row_heading level0 row6\" >6</th>\n",
       "      <td id=\"T_78eee_row6_col0\" class=\"data row6 col0\" >0.013000</td>\n",
       "      <td id=\"T_78eee_row6_col1\" class=\"data row6 col1\" >0.715714</td>\n",
       "      <td id=\"T_78eee_row6_col2\" class=\"data row6 col2\" >-0.085812</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row7\" class=\"row_heading level0 row7\" >7</th>\n",
       "      <td id=\"T_78eee_row7_col0\" class=\"data row7 col0\" >0.015000</td>\n",
       "      <td id=\"T_78eee_row7_col1\" class=\"data row7 col1\" >0.715714</td>\n",
       "      <td id=\"T_78eee_row7_col2\" class=\"data row7 col2\" >-0.085812</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row8\" class=\"row_heading level0 row8\" >8</th>\n",
       "      <td id=\"T_78eee_row8_col0\" class=\"data row8 col0\" >0.017000</td>\n",
       "      <td id=\"T_78eee_row8_col1\" class=\"data row8 col1\" >0.717143</td>\n",
       "      <td id=\"T_78eee_row8_col2\" class=\"data row8 col2\" >-0.078669</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row9\" class=\"row_heading level0 row9\" >9</th>\n",
       "      <td id=\"T_78eee_row9_col0\" class=\"data row9 col0\" >0.019000</td>\n",
       "      <td id=\"T_78eee_row9_col1\" class=\"data row9 col1\" >0.717143</td>\n",
       "      <td id=\"T_78eee_row9_col2\" class=\"data row9 col2\" >-0.078669</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row10\" class=\"row_heading level0 row10\" >10</th>\n",
       "      <td id=\"T_78eee_row10_col0\" class=\"data row10 col0\" >0.021000</td>\n",
       "      <td id=\"T_78eee_row10_col1\" class=\"data row10 col1\" >0.715714</td>\n",
       "      <td id=\"T_78eee_row10_col2\" class=\"data row10 col2\" >-0.071526</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row11\" class=\"row_heading level0 row11\" >11</th>\n",
       "      <td id=\"T_78eee_row11_col0\" class=\"data row11 col0\" >0.023000</td>\n",
       "      <td id=\"T_78eee_row11_col1\" class=\"data row11 col1\" >0.715714</td>\n",
       "      <td id=\"T_78eee_row11_col2\" class=\"data row11 col2\" >-0.071526</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row12\" class=\"row_heading level0 row12\" >12</th>\n",
       "      <td id=\"T_78eee_row12_col0\" class=\"data row12 col0\" >0.025000</td>\n",
       "      <td id=\"T_78eee_row12_col1\" class=\"data row12 col1\" >0.715714</td>\n",
       "      <td id=\"T_78eee_row12_col2\" class=\"data row12 col2\" >-0.071526</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row13\" class=\"row_heading level0 row13\" >13</th>\n",
       "      <td id=\"T_78eee_row13_col0\" class=\"data row13 col0\" >0.027000</td>\n",
       "      <td id=\"T_78eee_row13_col1\" class=\"data row13 col1\" >0.715714</td>\n",
       "      <td id=\"T_78eee_row13_col2\" class=\"data row13 col2\" >-0.071526</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row14\" class=\"row_heading level0 row14\" >14</th>\n",
       "      <td id=\"T_78eee_row14_col0\" class=\"data row14 col0\" >0.029000</td>\n",
       "      <td id=\"T_78eee_row14_col1\" class=\"data row14 col1\" >0.715714</td>\n",
       "      <td id=\"T_78eee_row14_col2\" class=\"data row14 col2\" >-0.071526</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row15\" class=\"row_heading level0 row15\" >15</th>\n",
       "      <td id=\"T_78eee_row15_col0\" class=\"data row15 col0\" >0.031000</td>\n",
       "      <td id=\"T_78eee_row15_col1\" class=\"data row15 col1\" >0.715714</td>\n",
       "      <td id=\"T_78eee_row15_col2\" class=\"data row15 col2\" >-0.071526</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row16\" class=\"row_heading level0 row16\" >16</th>\n",
       "      <td id=\"T_78eee_row16_col0\" class=\"data row16 col0\" >0.033000</td>\n",
       "      <td id=\"T_78eee_row16_col1\" class=\"data row16 col1\" >0.715714</td>\n",
       "      <td id=\"T_78eee_row16_col2\" class=\"data row16 col2\" >-0.071526</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row17\" class=\"row_heading level0 row17\" >17</th>\n",
       "      <td id=\"T_78eee_row17_col0\" class=\"data row17 col0\" >0.035000</td>\n",
       "      <td id=\"T_78eee_row17_col1\" class=\"data row17 col1\" >0.715714</td>\n",
       "      <td id=\"T_78eee_row17_col2\" class=\"data row17 col2\" >-0.071526</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row18\" class=\"row_heading level0 row18\" >18</th>\n",
       "      <td id=\"T_78eee_row18_col0\" class=\"data row18 col0\" >0.037000</td>\n",
       "      <td id=\"T_78eee_row18_col1\" class=\"data row18 col1\" >0.715714</td>\n",
       "      <td id=\"T_78eee_row18_col2\" class=\"data row18 col2\" >-0.071526</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row19\" class=\"row_heading level0 row19\" >19</th>\n",
       "      <td id=\"T_78eee_row19_col0\" class=\"data row19 col0\" >0.039000</td>\n",
       "      <td id=\"T_78eee_row19_col1\" class=\"data row19 col1\" >0.715714</td>\n",
       "      <td id=\"T_78eee_row19_col2\" class=\"data row19 col2\" >-0.071526</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row20\" class=\"row_heading level0 row20\" >20</th>\n",
       "      <td id=\"T_78eee_row20_col0\" class=\"data row20 col0\" >0.041000</td>\n",
       "      <td id=\"T_78eee_row20_col1\" class=\"data row20 col1\" >0.715714</td>\n",
       "      <td id=\"T_78eee_row20_col2\" class=\"data row20 col2\" >-0.071526</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row21\" class=\"row_heading level0 row21\" >21</th>\n",
       "      <td id=\"T_78eee_row21_col0\" class=\"data row21 col0\" >0.043000</td>\n",
       "      <td id=\"T_78eee_row21_col1\" class=\"data row21 col1\" >0.715714</td>\n",
       "      <td id=\"T_78eee_row21_col2\" class=\"data row21 col2\" >-0.071526</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row22\" class=\"row_heading level0 row22\" >22</th>\n",
       "      <td id=\"T_78eee_row22_col0\" class=\"data row22 col0\" >0.045000</td>\n",
       "      <td id=\"T_78eee_row22_col1\" class=\"data row22 col1\" >0.715714</td>\n",
       "      <td id=\"T_78eee_row22_col2\" class=\"data row22 col2\" >-0.071526</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row23\" class=\"row_heading level0 row23\" >23</th>\n",
       "      <td id=\"T_78eee_row23_col0\" class=\"data row23 col0\" >0.047000</td>\n",
       "      <td id=\"T_78eee_row23_col1\" class=\"data row23 col1\" >0.715714</td>\n",
       "      <td id=\"T_78eee_row23_col2\" class=\"data row23 col2\" >-0.071526</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row24\" class=\"row_heading level0 row24\" >24</th>\n",
       "      <td id=\"T_78eee_row24_col0\" class=\"data row24 col0\" >0.049000</td>\n",
       "      <td id=\"T_78eee_row24_col1\" class=\"data row24 col1\" >0.715714</td>\n",
       "      <td id=\"T_78eee_row24_col2\" class=\"data row24 col2\" >-0.071526</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row25\" class=\"row_heading level0 row25\" >25</th>\n",
       "      <td id=\"T_78eee_row25_col0\" class=\"data row25 col0\" >0.051000</td>\n",
       "      <td id=\"T_78eee_row25_col1\" class=\"data row25 col1\" >0.715714</td>\n",
       "      <td id=\"T_78eee_row25_col2\" class=\"data row25 col2\" >-0.050098</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row26\" class=\"row_heading level0 row26\" >26</th>\n",
       "      <td id=\"T_78eee_row26_col0\" class=\"data row26 col0\" >0.053000</td>\n",
       "      <td id=\"T_78eee_row26_col1\" class=\"data row26 col1\" >0.715714</td>\n",
       "      <td id=\"T_78eee_row26_col2\" class=\"data row26 col2\" >-0.050098</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row27\" class=\"row_heading level0 row27\" >27</th>\n",
       "      <td id=\"T_78eee_row27_col0\" class=\"data row27 col0\" >0.055000</td>\n",
       "      <td id=\"T_78eee_row27_col1\" class=\"data row27 col1\" >0.715714</td>\n",
       "      <td id=\"T_78eee_row27_col2\" class=\"data row27 col2\" >-0.050098</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row28\" class=\"row_heading level0 row28\" >28</th>\n",
       "      <td id=\"T_78eee_row28_col0\" class=\"data row28 col0\" >0.057000</td>\n",
       "      <td id=\"T_78eee_row28_col1\" class=\"data row28 col1\" >0.715714</td>\n",
       "      <td id=\"T_78eee_row28_col2\" class=\"data row28 col2\" >-0.050098</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row29\" class=\"row_heading level0 row29\" >29</th>\n",
       "      <td id=\"T_78eee_row29_col0\" class=\"data row29 col0\" >0.059000</td>\n",
       "      <td id=\"T_78eee_row29_col1\" class=\"data row29 col1\" >0.715714</td>\n",
       "      <td id=\"T_78eee_row29_col2\" class=\"data row29 col2\" >-0.050098</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row30\" class=\"row_heading level0 row30\" >30</th>\n",
       "      <td id=\"T_78eee_row30_col0\" class=\"data row30 col0\" >0.061000</td>\n",
       "      <td id=\"T_78eee_row30_col1\" class=\"data row30 col1\" >0.715714</td>\n",
       "      <td id=\"T_78eee_row30_col2\" class=\"data row30 col2\" >-0.050098</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row31\" class=\"row_heading level0 row31\" >31</th>\n",
       "      <td id=\"T_78eee_row31_col0\" class=\"data row31 col0\" >0.063000</td>\n",
       "      <td id=\"T_78eee_row31_col1\" class=\"data row31 col1\" >0.715714</td>\n",
       "      <td id=\"T_78eee_row31_col2\" class=\"data row31 col2\" >-0.050098</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row32\" class=\"row_heading level0 row32\" >32</th>\n",
       "      <td id=\"T_78eee_row32_col0\" class=\"data row32 col0\" >0.065000</td>\n",
       "      <td id=\"T_78eee_row32_col1\" class=\"data row32 col1\" >0.715714</td>\n",
       "      <td id=\"T_78eee_row32_col2\" class=\"data row32 col2\" >-0.050098</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row33\" class=\"row_heading level0 row33\" >33</th>\n",
       "      <td id=\"T_78eee_row33_col0\" class=\"data row33 col0\" >0.067000</td>\n",
       "      <td id=\"T_78eee_row33_col1\" class=\"data row33 col1\" >0.715714</td>\n",
       "      <td id=\"T_78eee_row33_col2\" class=\"data row33 col2\" >-0.050098</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_78eee_level0_row34\" class=\"row_heading level0 row34\" >34</th>\n",
       "      <td id=\"T_78eee_row34_col0\" class=\"data row34 col0\" >0.069000</td>\n",
       "      <td id=\"T_78eee_row34_col1\" class=\"data row34 col1\" >0.715714</td>\n",
       "      <td id=\"T_78eee_row34_col2\" class=\"data row34 col2\" >-0.050098</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n"
      ],
      "text/plain": [
       "<pandas.io.formats.style.Styler at 0x7f32c69b0390>"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#highlighting the best c value for fairness\n",
    "def syling(s):\n",
    "  if s['C'] == 0.051000000000000004:\n",
    "    return ['background-color: lightgreen']*3\n",
    "  else:\n",
    "    return ['background-color: white']*3\n",
    "\n",
    "dfg.style.apply(syling, axis = 1)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "sePZ1ypXUN4G"
   },
   "source": [
    "## Compute and Report Final Accuracy and Fairness Metric"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "HVlVQysC_o4Y"
   },
   "source": [
    "\n",
    "### Task 1: Table of Results from Training Data\n",
    "\n",
    "| Dataset | C Value for Highest Accuracy (1) | C Value for Best Fairness Metrics (2) |\n",
    "| --- | :---: | ---: |\n",
    "| **Adult** | 0.017 | 0.001 |\n",
    "| **German** | 0.009 | 0.051 |"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "VKAc2UXlIr0-"
   },
   "source": [
    "### Functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "id": "fphMNOv6ALE1"
   },
   "outputs": [],
   "source": [
    "#function for final scores\n",
    "\n",
    "def finalscores(X_train, y_train, X_test, y_test, test, c):\n",
    "\n",
    "  #empty lists\n",
    "  equ_opp_list = []\n",
    "  acc = []\n",
    "  disparate = []\n",
    "\n",
    "  #tarining the log red model\n",
    "  clf = LogisticRegression(solver='liblinear', C=c, random_state=1)\n",
    "  #fitting to the train values\n",
    "  clf.fit(X_train,y_train)\n",
    "  #predicting on the test set\n",
    "  y_pred = clf.predict(X_test)\n",
    "  test_pred = test.copy()\n",
    "  test_pred.labels = y_pred\n",
    "  \n",
    "  #appending the lists with accuracy scores\n",
    "  acc.append(sum(y_pred==y_test)/len(y_test))\n",
    "\n",
    "  #classifcation metric\n",
    "  metric = ClassificationMetric(test, test_pred, unprivileged_groups=unprivileged_groups, privileged_groups=privileged_groups)\n",
    "  #appending the lists with equ of opp and disparate impact\n",
    "  equ_opp_list.append(metric.equal_opportunity_difference()) #would like it to be close to 0\n",
    "  disparate.append(metric.disparate_impact()) #value close to 0 indicates strong bias\n",
    "\n",
    "  print('The final Accuracy score on the test data with hyperparameter C:', c, 'is', clf.score(X_test, y_test))\n",
    "  print('The final Equality of Opportunity score at C:', c, 'is', equ_opp_list)\n",
    "  print('The disparate impact score is', disparate)\n",
    "  "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "eK_jzKbh9xoX"
   },
   "source": [
    "### Adult\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "djsyyXv4-pP_"
   },
   "source": [
    "**Adult Dataset:** \n",
    "\n",
    "<font color='blue'>**(1)**</font> The hyperparameter for the highest accuracy on training data is at **C: 0.017**\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "9bzbmQvu-UnG",
    "outputId": "68abd462-66c1-450b-8ab7-99fde820f8fc"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The final Accuracy score on the test data with hyperparameter C: 0.017 is 0.8039991810550741\n",
      "The final Equality of Opportunity score at C: 0.017 is [-0.4581379772270596]\n",
      "The disparate impact score is [0.0]\n"
     ]
    }
   ],
   "source": [
    "#final scores for the adult test set, with hyperparameter 0.17\n",
    "\n",
    "finalscores(X_train_a, y_train_a, X_test_a, y_test_a, a_test, 0.017)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "DpaTCIeKGM09"
   },
   "source": [
    "> The <font color='blue'>**final *accuracy* score**</font> in testing data at hyperparameter C: 0.017 is <font color='blue'>**0.804**</font>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "T3LDvteTFeoK"
   },
   "source": [
    "<font color='green'>**(2)**</font> The hyperparameter for the best fairness metric on training data is at **C: 0.001**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "ACCxVmHECiNx",
    "outputId": "874d93e6-7791-4361-d856-1207058afe36"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The final Accuracy score on the test data with hyperparameter C: 0.001 is 0.8042721626970586\n",
      "The final Equality of Opportunity score at C: 0.001 is [-0.4511051574012056]\n",
      "The disparate impact score is [0.0]\n"
     ]
    }
   ],
   "source": [
    "#final scores for the adult test set, with hyperparameter 0.001\n",
    "\n",
    "finalscores(X_train_a, y_train_a, X_test_a, y_test_a, a_test, 0.001)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "f7TBSAKJKl0t"
   },
   "source": [
    "> The <font color='green'>**final *Equality of Opportunity* score**</font> in testing data at hyperparameter C: 0.001 is <font color='green'>**-0.45110**</font>\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "baH6dfQ-90e9"
   },
   "source": [
    "### German\n",
    "\n",
    "**German Dataset:** \n",
    "\n",
    "<font color='blue'>**(1)**</font>  The hyperparameter for the highest accuracy on training data is at **C: 0.009**\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "O-KLy0Rm_lJg",
    "outputId": "e3006d43-c67c-4c4e-922f-e9265b25c014"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The final Accuracy score on the test data with hyperparameter C: 0.009 is 0.6466666666666666\n",
      "The final Equality of Opportunity score at C: 0.009 is [-0.1504771225838023]\n",
      "The disparate impact score is [0.7754407051282051]\n"
     ]
    }
   ],
   "source": [
    "#final scores for the german test set, with hyperparameter 0.009\n",
    "\n",
    "finalscores(X_train_g, y_train_g, X_test_g, y_test_g, g_test, 0.009)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "4Fr2jFn0Fuc7"
   },
   "source": [
    "> The <font color='blue'>**final *accuracy* score**</font> in testing data at hyperparameter C: 0.009 is <font color='blue'>**0.6467**</font>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "tDWdM4dZFbyF"
   },
   "source": [
    "<font color='green'>**(2)**</font> The hyperparameter for best fairness metricon training data is at **C: 0.051**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "2sOzk3qTEOTj",
    "outputId": "37943835-553e-4d84-9e11-cc6f2e555e59"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The final Accuracy score on the test data with hyperparameter C: 0.051 is 0.6466666666666666\n",
      "The final Equality of Opportunity score at C: 0.051 is [-0.1504771225838023]\n",
      "The disparate impact score is [0.7754407051282051]\n"
     ]
    }
   ],
   "source": [
    "#final scores for the german test set, with hyperparameter 0.051\n",
    "\n",
    "finalscores(X_train_g, y_train_g, X_test_g, y_test_g, g_test, 0.051)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "GNZDgwFDLCgK"
   },
   "source": [
    "\n",
    "> The <font color='green'>**final *Equality of Opportunity* score**</font> on testing data at hyperparameter C: 0.051 is <font color='green'>**-0.1505**</font>\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "MxIABBhN0E4s"
   },
   "source": [
    "___"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "XduX0Sc1Gs4y"
   },
   "source": [
    "# **Task 2** \n",
    "\n",
    "Now choose an algorithmic fairness method, e.g. reweighing (Lecture 7), and perform\n",
    "the same analysis, i.e. how varying the hyperparameter(s) impacts accuracy and fairness metrics\n",
    "during model selection.\n",
    "Specifically, using the same 5-fold cross validation as in Task 1, perform validation by varying\n",
    "the trade-off hyperparameter and select (3) the model with the highest accuracy, and (4) the\n",
    "model with the best fairness metric across 5 folds. After model selection, compute and report\n",
    "final accuracy and fairness for both models (3) and (4) on the test set."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "fjmbewXAdrsb"
   },
   "source": [
    "## (3) Hyperparameter tuning using GridSearch: Re-weighing "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "grgkMwxoeFt7"
   },
   "source": [
    "### Functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "id": "z-eJPtt8ALlb"
   },
   "outputs": [],
   "source": [
    "#reweighing gridsearch function\n",
    "\n",
    "def rw_gridsearch (X, y, train,cvals):\n",
    "  #mitgating the bias\n",
    "  #assigning the new weights to the unprivileged and privileged groups, transform the dataset with reweighing\n",
    "  RW = Reweighing(unprivileged_groups=unprivileged_groups,\n",
    "                privileged_groups=privileged_groups)\n",
    "  #obtaining a set of weights for the training set\n",
    "  train = RW.fit_transform(train)\n",
    "  #training the model and using gridsearch, 5-fold cross validation\n",
    "  clf= GridSearchCV(LogisticRegression(solver = 'liblinear', random_state = 1), {'C':cvals}, cv=5)\n",
    "  #fitting to new weights\n",
    "  clf.fit(X,y,sample_weight=train.instance_weights)\n",
    "\n",
    "  #displaying the results\n",
    "  clf.cv_results_\n",
    "  print('Tuned hyperparameters : ',clf.best_params_, 'with accuracy of', clf.best_score_)\n",
    "\n",
    "  #showing the dataframe\n",
    "  df = pd.DataFrame(clf.cv_results_)\n",
    "  display(df[['params','mean_test_score' ]])\n",
    "\n",
    "  #plotting the results\n",
    "  xplt = np.array(cvals)\n",
    "  plt.figure(figsize = (10,6))\n",
    "  plt.plot(xplt, df['mean_test_score' ], color = 'purple')\n",
    "  font1 = {'family':'serif','color':'purple','size':13}\n",
    "  font2 = {'family':'serif','color':'purple','size':11}\n",
    "  plt.title('C Value vs Average', fontdict = font1)\n",
    "  plt.xlabel('C Value', fontdict = font2)\n",
    "  plt.ylabel('Average', fontdict = font2)\n",
    "  plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "IHKnYWH8d8S6"
   },
   "source": [
    "### Adult"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 642
    },
    "id": "eRTaFPT6Rhl3",
    "outputId": "dd167719-c457-4e2a-dba0-7575fbeea266"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tuned hyperparameters :  {'C': 0.05} with accuracy of 0.7884409465674679\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "\n",
       "  <div id=\"df-79c820ba-2617-42df-8848-38f9816c13f8\">\n",
       "    <div class=\"colab-df-container\">\n",
       "      <div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>params</th>\n",
       "      <th>mean_test_score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>{'C': 0.005}</td>\n",
       "      <td>0.788382</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>{'C': 0.01}</td>\n",
       "      <td>0.787944</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>{'C': 0.05}</td>\n",
       "      <td>0.788441</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>{'C': 0.1}</td>\n",
       "      <td>0.788441</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>{'C': 0.5}</td>\n",
       "      <td>0.788441</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>{'C': 1}</td>\n",
       "      <td>0.788441</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>\n",
       "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-79c820ba-2617-42df-8848-38f9816c13f8')\"\n",
       "              title=\"Convert this dataframe to an interactive table.\"\n",
       "              style=\"display:none;\">\n",
       "        \n",
       "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
       "       width=\"24px\">\n",
       "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
       "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
       "  </svg>\n",
       "      </button>\n",
       "      \n",
       "  <style>\n",
       "    .colab-df-container {\n",
       "      display:flex;\n",
       "      flex-wrap:wrap;\n",
       "      gap: 12px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert {\n",
       "      background-color: #E8F0FE;\n",
       "      border: none;\n",
       "      border-radius: 50%;\n",
       "      cursor: pointer;\n",
       "      display: none;\n",
       "      fill: #1967D2;\n",
       "      height: 32px;\n",
       "      padding: 0 0 0 0;\n",
       "      width: 32px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert:hover {\n",
       "      background-color: #E2EBFA;\n",
       "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
       "      fill: #174EA6;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert {\n",
       "      background-color: #3B4455;\n",
       "      fill: #D2E3FC;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert:hover {\n",
       "      background-color: #434B5C;\n",
       "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
       "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
       "      fill: #FFFFFF;\n",
       "    }\n",
       "  </style>\n",
       "\n",
       "      <script>\n",
       "        const buttonEl =\n",
       "          document.querySelector('#df-79c820ba-2617-42df-8848-38f9816c13f8 button.colab-df-convert');\n",
       "        buttonEl.style.display =\n",
       "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
       "\n",
       "        async function convertToInteractive(key) {\n",
       "          const element = document.querySelector('#df-79c820ba-2617-42df-8848-38f9816c13f8');\n",
       "          const dataTable =\n",
       "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
       "                                                     [key], {});\n",
       "          if (!dataTable) return;\n",
       "\n",
       "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
       "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
       "            + ' to learn more about interactive tables.';\n",
       "          element.innerHTML = '';\n",
       "          dataTable['output_type'] = 'display_data';\n",
       "          await google.colab.output.renderOutput(dataTable, element);\n",
       "          const docLink = document.createElement('div');\n",
       "          docLink.innerHTML = docLinkHtml;\n",
       "          element.appendChild(docLink);\n",
       "        }\n",
       "      </script>\n",
       "    </div>\n",
       "  </div>\n",
       "  "
      ],
      "text/plain": [
       "         params  mean_test_score\n",
       "0  {'C': 0.005}         0.788382\n",
       "1   {'C': 0.01}         0.787944\n",
       "2   {'C': 0.05}         0.788441\n",
       "3    {'C': 0.1}         0.788441\n",
       "4    {'C': 0.5}         0.788441\n",
       "5      {'C': 1}         0.788441"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#reweigh the adult dataset and perfrom gridsearch\n",
    "#on a broad range of c\n",
    "\n",
    "rw_gridsearch (X_train_a, y_train_a, a_train, c)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "id": "r4Y24snGHph5",
    "outputId": "f7d24149-dd42-4ef1-d9d6-30d9ebe2699d"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tuned hyperparameters :  {'C': 0.001} with accuracy of 0.7891137006660291\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "\n",
       "  <div id=\"df-a170ff64-f457-4315-b6a9-8b64720e9dbd\">\n",
       "    <div class=\"colab-df-container\">\n",
       "      <div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>params</th>\n",
       "      <th>mean_test_score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>{'C': 0.001}</td>\n",
       "      <td>0.789114</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>{'C': 0.003}</td>\n",
       "      <td>0.788704</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>{'C': 0.005}</td>\n",
       "      <td>0.788382</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>{'C': 0.007}</td>\n",
       "      <td>0.788382</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>{'C': 0.009000000000000001}</td>\n",
       "      <td>0.787944</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>{'C': 0.011}</td>\n",
       "      <td>0.788002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>{'C': 0.013000000000000001}</td>\n",
       "      <td>0.788002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>{'C': 0.015}</td>\n",
       "      <td>0.788441</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>{'C': 0.017}</td>\n",
       "      <td>0.788441</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>{'C': 0.019000000000000003}</td>\n",
       "      <td>0.788441</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>{'C': 0.021}</td>\n",
       "      <td>0.788441</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>{'C': 0.023}</td>\n",
       "      <td>0.788441</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>{'C': 0.025}</td>\n",
       "      <td>0.788441</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>{'C': 0.027000000000000003}</td>\n",
       "      <td>0.788441</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>{'C': 0.029}</td>\n",
       "      <td>0.788441</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>{'C': 0.031}</td>\n",
       "      <td>0.788441</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>{'C': 0.033}</td>\n",
       "      <td>0.788441</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>{'C': 0.035}</td>\n",
       "      <td>0.788441</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>{'C': 0.037000000000000005}</td>\n",
       "      <td>0.788441</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>{'C': 0.039}</td>\n",
       "      <td>0.788441</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>{'C': 0.041}</td>\n",
       "      <td>0.788441</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>{'C': 0.043000000000000003}</td>\n",
       "      <td>0.788441</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>{'C': 0.045}</td>\n",
       "      <td>0.788441</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>{'C': 0.047}</td>\n",
       "      <td>0.788441</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>{'C': 0.049}</td>\n",
       "      <td>0.788441</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>{'C': 0.051000000000000004}</td>\n",
       "      <td>0.788441</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>{'C': 0.053000000000000005}</td>\n",
       "      <td>0.788441</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>{'C': 0.055}</td>\n",
       "      <td>0.788441</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>{'C': 0.057}</td>\n",
       "      <td>0.788441</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>{'C': 0.059000000000000004}</td>\n",
       "      <td>0.788441</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>{'C': 0.061}</td>\n",
       "      <td>0.788441</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>31</th>\n",
       "      <td>{'C': 0.063}</td>\n",
       "      <td>0.788441</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32</th>\n",
       "      <td>{'C': 0.065}</td>\n",
       "      <td>0.788441</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33</th>\n",
       "      <td>{'C': 0.067}</td>\n",
       "      <td>0.788441</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>34</th>\n",
       "      <td>{'C': 0.069}</td>\n",
       "      <td>0.788441</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>\n",
       "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-a170ff64-f457-4315-b6a9-8b64720e9dbd')\"\n",
       "              title=\"Convert this dataframe to an interactive table.\"\n",
       "              style=\"display:none;\">\n",
       "        \n",
       "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
       "       width=\"24px\">\n",
       "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
       "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
       "  </svg>\n",
       "      </button>\n",
       "      \n",
       "  <style>\n",
       "    .colab-df-container {\n",
       "      display:flex;\n",
       "      flex-wrap:wrap;\n",
       "      gap: 12px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert {\n",
       "      background-color: #E8F0FE;\n",
       "      border: none;\n",
       "      border-radius: 50%;\n",
       "      cursor: pointer;\n",
       "      display: none;\n",
       "      fill: #1967D2;\n",
       "      height: 32px;\n",
       "      padding: 0 0 0 0;\n",
       "      width: 32px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert:hover {\n",
       "      background-color: #E2EBFA;\n",
       "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
       "      fill: #174EA6;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert {\n",
       "      background-color: #3B4455;\n",
       "      fill: #D2E3FC;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert:hover {\n",
       "      background-color: #434B5C;\n",
       "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
       "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
       "      fill: #FFFFFF;\n",
       "    }\n",
       "  </style>\n",
       "\n",
       "      <script>\n",
       "        const buttonEl =\n",
       "          document.querySelector('#df-a170ff64-f457-4315-b6a9-8b64720e9dbd button.colab-df-convert');\n",
       "        buttonEl.style.display =\n",
       "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
       "\n",
       "        async function convertToInteractive(key) {\n",
       "          const element = document.querySelector('#df-a170ff64-f457-4315-b6a9-8b64720e9dbd');\n",
       "          const dataTable =\n",
       "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
       "                                                     [key], {});\n",
       "          if (!dataTable) return;\n",
       "\n",
       "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
       "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
       "            + ' to learn more about interactive tables.';\n",
       "          element.innerHTML = '';\n",
       "          dataTable['output_type'] = 'display_data';\n",
       "          await google.colab.output.renderOutput(dataTable, element);\n",
       "          const docLink = document.createElement('div');\n",
       "          docLink.innerHTML = docLinkHtml;\n",
       "          element.appendChild(docLink);\n",
       "        }\n",
       "      </script>\n",
       "    </div>\n",
       "  </div>\n",
       "  "
      ],
      "text/plain": [
       "                         params  mean_test_score\n",
       "0                  {'C': 0.001}         0.789114\n",
       "1                  {'C': 0.003}         0.788704\n",
       "2                  {'C': 0.005}         0.788382\n",
       "3                  {'C': 0.007}         0.788382\n",
       "4   {'C': 0.009000000000000001}         0.787944\n",
       "5                  {'C': 0.011}         0.788002\n",
       "6   {'C': 0.013000000000000001}         0.788002\n",
       "7                  {'C': 0.015}         0.788441\n",
       "8                  {'C': 0.017}         0.788441\n",
       "9   {'C': 0.019000000000000003}         0.788441\n",
       "10                 {'C': 0.021}         0.788441\n",
       "11                 {'C': 0.023}         0.788441\n",
       "12                 {'C': 0.025}         0.788441\n",
       "13  {'C': 0.027000000000000003}         0.788441\n",
       "14                 {'C': 0.029}         0.788441\n",
       "15                 {'C': 0.031}         0.788441\n",
       "16                 {'C': 0.033}         0.788441\n",
       "17                 {'C': 0.035}         0.788441\n",
       "18  {'C': 0.037000000000000005}         0.788441\n",
       "19                 {'C': 0.039}         0.788441\n",
       "20                 {'C': 0.041}         0.788441\n",
       "21  {'C': 0.043000000000000003}         0.788441\n",
       "22                 {'C': 0.045}         0.788441\n",
       "23                 {'C': 0.047}         0.788441\n",
       "24                 {'C': 0.049}         0.788441\n",
       "25  {'C': 0.051000000000000004}         0.788441\n",
       "26  {'C': 0.053000000000000005}         0.788441\n",
       "27                 {'C': 0.055}         0.788441\n",
       "28                 {'C': 0.057}         0.788441\n",
       "29  {'C': 0.059000000000000004}         0.788441\n",
       "30                 {'C': 0.061}         0.788441\n",
       "31                 {'C': 0.063}         0.788441\n",
       "32                 {'C': 0.065}         0.788441\n",
       "33                 {'C': 0.067}         0.788441\n",
       "34                 {'C': 0.069}         0.788441"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#refining into c more\n",
    "\n",
    "rw_gridsearch(X_train_a, y_train_a, a_train, np.arange(0.001, 0.07, 0.002))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "e3hEJaxNNitN"
   },
   "source": [
    "**Adult Dataset:** The hyperparameter for the highest accuracy (**0.7891**) is at **C: 0.001**\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "xcf39qpqd-l7"
   },
   "source": [
    "### German"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 642
    },
    "id": "taQZAxknMY3q",
    "outputId": "1f468ec2-7dc4-4033-c50e-d6b1f8fb3c98"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tuned hyperparameters :  {'C': 0.5} with accuracy of 0.7128571428571429\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "\n",
       "  <div id=\"df-68c735ef-f1ca-4edb-b921-72c67a47c2fe\">\n",
       "    <div class=\"colab-df-container\">\n",
       "      <div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>params</th>\n",
       "      <th>mean_test_score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>{'C': 0.005}</td>\n",
       "      <td>0.711429</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>{'C': 0.01}</td>\n",
       "      <td>0.705714</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>{'C': 0.05}</td>\n",
       "      <td>0.708571</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>{'C': 0.1}</td>\n",
       "      <td>0.710000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>{'C': 0.5}</td>\n",
       "      <td>0.712857</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>{'C': 1}</td>\n",
       "      <td>0.712857</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>\n",
       "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-68c735ef-f1ca-4edb-b921-72c67a47c2fe')\"\n",
       "              title=\"Convert this dataframe to an interactive table.\"\n",
       "              style=\"display:none;\">\n",
       "        \n",
       "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
       "       width=\"24px\">\n",
       "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
       "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
       "  </svg>\n",
       "      </button>\n",
       "      \n",
       "  <style>\n",
       "    .colab-df-container {\n",
       "      display:flex;\n",
       "      flex-wrap:wrap;\n",
       "      gap: 12px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert {\n",
       "      background-color: #E8F0FE;\n",
       "      border: none;\n",
       "      border-radius: 50%;\n",
       "      cursor: pointer;\n",
       "      display: none;\n",
       "      fill: #1967D2;\n",
       "      height: 32px;\n",
       "      padding: 0 0 0 0;\n",
       "      width: 32px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert:hover {\n",
       "      background-color: #E2EBFA;\n",
       "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
       "      fill: #174EA6;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert {\n",
       "      background-color: #3B4455;\n",
       "      fill: #D2E3FC;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert:hover {\n",
       "      background-color: #434B5C;\n",
       "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
       "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
       "      fill: #FFFFFF;\n",
       "    }\n",
       "  </style>\n",
       "\n",
       "      <script>\n",
       "        const buttonEl =\n",
       "          document.querySelector('#df-68c735ef-f1ca-4edb-b921-72c67a47c2fe button.colab-df-convert');\n",
       "        buttonEl.style.display =\n",
       "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
       "\n",
       "        async function convertToInteractive(key) {\n",
       "          const element = document.querySelector('#df-68c735ef-f1ca-4edb-b921-72c67a47c2fe');\n",
       "          const dataTable =\n",
       "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
       "                                                     [key], {});\n",
       "          if (!dataTable) return;\n",
       "\n",
       "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
       "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
       "            + ' to learn more about interactive tables.';\n",
       "          element.innerHTML = '';\n",
       "          dataTable['output_type'] = 'display_data';\n",
       "          await google.colab.output.renderOutput(dataTable, element);\n",
       "          const docLink = document.createElement('div');\n",
       "          docLink.innerHTML = docLinkHtml;\n",
       "          element.appendChild(docLink);\n",
       "        }\n",
       "      </script>\n",
       "    </div>\n",
       "  </div>\n",
       "  "
      ],
      "text/plain": [
       "         params  mean_test_score\n",
       "0  {'C': 0.005}         0.711429\n",
       "1   {'C': 0.01}         0.705714\n",
       "2   {'C': 0.05}         0.708571\n",
       "3    {'C': 0.1}         0.710000\n",
       "4    {'C': 0.5}         0.712857\n",
       "5      {'C': 1}         0.712857"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#reweigh the german dataset and perfrom gridsearch\n",
    "#on a broad range of c\n",
    "\n",
    "rw_gridsearch (X_train_g, y_train_g, g_train, c)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "jpDXl2YKI9Ul"
   },
   "source": [
    "Expanding the C values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 828
    },
    "id": "6p6PSGjlMTbj",
    "outputId": "f09c33b6-2bc3-4e7d-f90a-f86615e3fc35"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tuned hyperparameters :  {'C': 0.10999999999999997} with accuracy of 0.7128571428571429\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "\n",
       "  <div id=\"df-dd27ad51-a3de-4bd7-826f-3e07a73dda16\">\n",
       "    <div class=\"colab-df-container\">\n",
       "      <div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>params</th>\n",
       "      <th>mean_test_score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>{'C': 0.01}</td>\n",
       "      <td>0.705714</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>{'C': 0.015}</td>\n",
       "      <td>0.710000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>{'C': 0.019999999999999997}</td>\n",
       "      <td>0.710000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>{'C': 0.024999999999999998}</td>\n",
       "      <td>0.707143</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>{'C': 0.03}</td>\n",
       "      <td>0.704286</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>193</th>\n",
       "      <td>{'C': 0.9749999999999999}</td>\n",
       "      <td>0.712857</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>194</th>\n",
       "      <td>{'C': 0.9799999999999999}</td>\n",
       "      <td>0.712857</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>195</th>\n",
       "      <td>{'C': 0.9849999999999999}</td>\n",
       "      <td>0.712857</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>196</th>\n",
       "      <td>{'C': 0.9899999999999999}</td>\n",
       "      <td>0.712857</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>197</th>\n",
       "      <td>{'C': 0.9949999999999999}</td>\n",
       "      <td>0.712857</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>198 rows × 2 columns</p>\n",
       "</div>\n",
       "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-dd27ad51-a3de-4bd7-826f-3e07a73dda16')\"\n",
       "              title=\"Convert this dataframe to an interactive table.\"\n",
       "              style=\"display:none;\">\n",
       "        \n",
       "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
       "       width=\"24px\">\n",
       "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
       "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
       "  </svg>\n",
       "      </button>\n",
       "      \n",
       "  <style>\n",
       "    .colab-df-container {\n",
       "      display:flex;\n",
       "      flex-wrap:wrap;\n",
       "      gap: 12px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert {\n",
       "      background-color: #E8F0FE;\n",
       "      border: none;\n",
       "      border-radius: 50%;\n",
       "      cursor: pointer;\n",
       "      display: none;\n",
       "      fill: #1967D2;\n",
       "      height: 32px;\n",
       "      padding: 0 0 0 0;\n",
       "      width: 32px;\n",
       "    }\n",
       "\n",
       "    .colab-df-convert:hover {\n",
       "      background-color: #E2EBFA;\n",
       "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
       "      fill: #174EA6;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert {\n",
       "      background-color: #3B4455;\n",
       "      fill: #D2E3FC;\n",
       "    }\n",
       "\n",
       "    [theme=dark] .colab-df-convert:hover {\n",
       "      background-color: #434B5C;\n",
       "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
       "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
       "      fill: #FFFFFF;\n",
       "    }\n",
       "  </style>\n",
       "\n",
       "      <script>\n",
       "        const buttonEl =\n",
       "          document.querySelector('#df-dd27ad51-a3de-4bd7-826f-3e07a73dda16 button.colab-df-convert');\n",
       "        buttonEl.style.display =\n",
       "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
       "\n",
       "        async function convertToInteractive(key) {\n",
       "          const element = document.querySelector('#df-dd27ad51-a3de-4bd7-826f-3e07a73dda16');\n",
       "          const dataTable =\n",
       "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
       "                                                     [key], {});\n",
       "          if (!dataTable) return;\n",
       "\n",
       "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
       "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
       "            + ' to learn more about interactive tables.';\n",
       "          element.innerHTML = '';\n",
       "          dataTable['output_type'] = 'display_data';\n",
       "          await google.colab.output.renderOutput(dataTable, element);\n",
       "          const docLink = document.createElement('div');\n",
       "          docLink.innerHTML = docLinkHtml;\n",
       "          element.appendChild(docLink);\n",
       "        }\n",
       "      </script>\n",
       "    </div>\n",
       "  </div>\n",
       "  "
      ],
      "text/plain": [
       "                          params  mean_test_score\n",
       "0                    {'C': 0.01}         0.705714\n",
       "1                   {'C': 0.015}         0.710000\n",
       "2    {'C': 0.019999999999999997}         0.710000\n",
       "3    {'C': 0.024999999999999998}         0.707143\n",
       "4                    {'C': 0.03}         0.704286\n",
       "..                           ...              ...\n",
       "193    {'C': 0.9749999999999999}         0.712857\n",
       "194    {'C': 0.9799999999999999}         0.712857\n",
       "195    {'C': 0.9849999999999999}         0.712857\n",
       "196    {'C': 0.9899999999999999}         0.712857\n",
       "197    {'C': 0.9949999999999999}         0.712857\n",
       "\n",
       "[198 rows x 2 columns]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#refining into c more\n",
    "\n",
    "rw_gridsearch (X_train_g, y_train_g, g_train, np.arange(0.01, 1, 0.005))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "tQICGNV_Ow8z"
   },
   "source": [
    "**German Dataset:** The hyperparameter for the highest accuracy (**0.7129**) is at **C: 0.12**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "JaIUVlgweh6M"
   },
   "source": [
    "## (4) Equality of Opportunity Fairness Metrics: Re-weighing "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "vjHus-3de7-L"
   },
   "source": [
    "### Functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "id": "XDJwWwEdRN1_"
   },
   "outputs": [],
   "source": [
    "def cv_rw_fairness(X_train, y_train, train, cval):\n",
    "  c_values = cval\n",
    "  #empty lists\n",
    "  acc_scores = []\n",
    "  fairness = []\n",
    "  equ_opp = []\n",
    "  avg_acc_scores = []\n",
    "\n",
    "  #mitgating the bias\n",
    "  #assigning the new weights to the unprivileged and privileged groups, transform the dataset with reweighing\n",
    "  RW = Reweighing(unprivileged_groups=unprivileged_groups,\n",
    "                privileged_groups=privileged_groups)\n",
    "  #obtaining a set of weights for the training set\n",
    "  train = RW.fit_transform(train)\n",
    "\n",
    "  \n",
    "  for i in c_values:\n",
    "    clf = LogisticRegression(solver = 'liblinear', C=i, random_state = 1)\n",
    "\n",
    "    #accuracy score with cross val score\n",
    "    #as reweighed, set params to 'sample weight': train.instance_weights\n",
    "    score = cross_val_score(clf, X_train, y_train, cv = 5, fit_params={'sample_weight': train.instance_weights}) #cv = 5 means 5 fold cross validation\n",
    "    #fairness predictions with cross val predict\n",
    "    #generates cross-validated estimates for each input data point\n",
    "    fairness = cross_val_predict(clf, X_train, y_train, cv = 5, fit_params={'sample_weight': train.instance_weights}) #cv = 5 means 5 fold cross validation\n",
    "    test_pred = train.copy()\n",
    "    #get the fair prediction labels\n",
    "    test_pred.labels = fairness\n",
    "\n",
    "    #defining our metric\n",
    "    metric = ClassificationMetric(train, test_pred, unprivileged_groups=unprivileged_groups, privileged_groups=privileged_groups)\n",
    "\n",
    "    #equality of opportunity score\n",
    "    equ_opp.append(metric.equal_opportunity_difference())\n",
    "\n",
    "    #average score\n",
    "    avg_acc = np.mean(score)\n",
    "    avg_acc_scores.append(np.mean(score))\n",
    "\n",
    "  \n",
    "  #defining the best \n",
    "  equal_opp_best = min(equ_opp, key= abs)\n",
    "  best_c = equ_opp.index(equal_opp_best)\n",
    "\n",
    "  #best average score\n",
    "  best_acc_c = avg_acc_scores.index(max(avg_acc_scores))\n",
    "  \n",
    "  #reporting the results with the best c value\n",
    "  print('Best accuracy score is at C: ', c_values[best_acc_c], 'with accuracy score of', max(avg_acc_scores))\n",
    "  print('Best Equality of Opportunity score is at C:',c_values[best_c],  'with score' ,equal_opp_best)\n",
    "\n",
    "  #reporting the results in a dataframe\n",
    "  df = pd.DataFrame({'C': c_values, \n",
    "                  'Accuracy' :avg_acc_scores,\n",
    "                  'Equality of Opportunity': equ_opp})\n",
    "  \n",
    "  #plotting results\n",
    "  xplt = np.array(cval)\n",
    "  plt.figure(figsize = (10,6))\n",
    "  plt.plot(xplt, df['Equality of Opportunity'], color = 'darkorange')\n",
    "  font1 = {'family':'serif','color':'darkorange','size':13}\n",
    "  font2 = {'family':'serif','color':'darkorange','size':11}\n",
    "  plt.title('Reweighing: Equality of Opportunity Hyperparameter tuning', fontdict = font1)\n",
    "  plt.xlabel('C Value', fontdict = font2)\n",
    "  plt.ylabel('Equality of Opportunity', fontdict = font2)\n",
    "  plt.show()\n",
    "  return(df)\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "96XCFv51fAEb"
   },
   "source": [
    "### Adult"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 439
    },
    "id": "p5tM6IFJUrUK",
    "outputId": "99af223b-b147-43a4-c74e-0f6e881a5c4f"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best accuracy score is at C:  0.001 with accuracy score of 0.7891137006660291\n",
      "Best Equality of Opportunity score is at C: 0.003 with score -0.008177456922513315\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#reweighted cross fold on adults dataset with refined c\n",
    "\n",
    "dfa3 = cv_rw_fairness(X_train_a, y_train_a, a_train, a)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "EfFY5dWbT9-E"
   },
   "source": [
    "**Adult Dataset:** (Re-weighing) The hyperparameter for the best *Equality of Opportunity* is at **C: 0.003** with a score of **-0.008017**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "id": "s9I6XbzyR80W",
    "outputId": "6d72aed7-7923-4425-8f98-78a49b6a9d0d"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style type=\"text/css\">\n",
       "#T_672bb_row0_col0, #T_672bb_row0_col1, #T_672bb_row0_col2, #T_672bb_row2_col0, #T_672bb_row2_col1, #T_672bb_row2_col2, #T_672bb_row3_col0, #T_672bb_row3_col1, #T_672bb_row3_col2, #T_672bb_row4_col0, #T_672bb_row4_col1, #T_672bb_row4_col2, #T_672bb_row5_col0, #T_672bb_row5_col1, #T_672bb_row5_col2, #T_672bb_row6_col0, #T_672bb_row6_col1, #T_672bb_row6_col2, #T_672bb_row7_col0, #T_672bb_row7_col1, #T_672bb_row7_col2, #T_672bb_row8_col0, #T_672bb_row8_col1, #T_672bb_row8_col2, #T_672bb_row9_col0, #T_672bb_row9_col1, #T_672bb_row9_col2, #T_672bb_row10_col0, #T_672bb_row10_col1, #T_672bb_row10_col2, #T_672bb_row11_col0, #T_672bb_row11_col1, #T_672bb_row11_col2, #T_672bb_row12_col0, #T_672bb_row12_col1, #T_672bb_row12_col2, #T_672bb_row13_col0, #T_672bb_row13_col1, #T_672bb_row13_col2, #T_672bb_row14_col0, #T_672bb_row14_col1, #T_672bb_row14_col2, #T_672bb_row15_col0, #T_672bb_row15_col1, #T_672bb_row15_col2, #T_672bb_row16_col0, #T_672bb_row16_col1, #T_672bb_row16_col2, #T_672bb_row17_col0, #T_672bb_row17_col1, #T_672bb_row17_col2, #T_672bb_row18_col0, #T_672bb_row18_col1, #T_672bb_row18_col2, #T_672bb_row19_col0, #T_672bb_row19_col1, #T_672bb_row19_col2, #T_672bb_row20_col0, #T_672bb_row20_col1, #T_672bb_row20_col2, #T_672bb_row21_col0, #T_672bb_row21_col1, #T_672bb_row21_col2, #T_672bb_row22_col0, #T_672bb_row22_col1, #T_672bb_row22_col2, #T_672bb_row23_col0, #T_672bb_row23_col1, #T_672bb_row23_col2, #T_672bb_row24_col0, #T_672bb_row24_col1, #T_672bb_row24_col2, #T_672bb_row25_col0, #T_672bb_row25_col1, #T_672bb_row25_col2, #T_672bb_row26_col0, #T_672bb_row26_col1, #T_672bb_row26_col2, #T_672bb_row27_col0, #T_672bb_row27_col1, #T_672bb_row27_col2, #T_672bb_row28_col0, #T_672bb_row28_col1, #T_672bb_row28_col2, #T_672bb_row29_col0, #T_672bb_row29_col1, #T_672bb_row29_col2, #T_672bb_row30_col0, #T_672bb_row30_col1, #T_672bb_row30_col2, #T_672bb_row31_col0, #T_672bb_row31_col1, #T_672bb_row31_col2, #T_672bb_row32_col0, #T_672bb_row32_col1, #T_672bb_row32_col2, #T_672bb_row33_col0, #T_672bb_row33_col1, #T_672bb_row33_col2, #T_672bb_row34_col0, #T_672bb_row34_col1, #T_672bb_row34_col2 {\n",
       "  background-color: white;\n",
       "}\n",
       "#T_672bb_row1_col0, #T_672bb_row1_col1, #T_672bb_row1_col2 {\n",
       "  background-color: orange;\n",
       "}\n",
       "</style>\n",
       "<table id=\"T_672bb_\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th class=\"blank level0\" >&nbsp;</th>\n",
       "      <th class=\"col_heading level0 col0\" >C</th>\n",
       "      <th class=\"col_heading level0 col1\" >Accuracy</th>\n",
       "      <th class=\"col_heading level0 col2\" >Equality of Opportunity</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row0\" class=\"row_heading level0 row0\" >0</th>\n",
       "      <td id=\"T_672bb_row0_col0\" class=\"data row0 col0\" >0.001000</td>\n",
       "      <td id=\"T_672bb_row0_col1\" class=\"data row0 col1\" >0.789114</td>\n",
       "      <td id=\"T_672bb_row0_col2\" class=\"data row0 col2\" >-0.012217</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row1\" class=\"row_heading level0 row1\" >1</th>\n",
       "      <td id=\"T_672bb_row1_col0\" class=\"data row1 col0\" >0.003000</td>\n",
       "      <td id=\"T_672bb_row1_col1\" class=\"data row1 col1\" >0.788704</td>\n",
       "      <td id=\"T_672bb_row1_col2\" class=\"data row1 col2\" >-0.008177</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row2\" class=\"row_heading level0 row2\" >2</th>\n",
       "      <td id=\"T_672bb_row2_col0\" class=\"data row2 col0\" >0.005000</td>\n",
       "      <td id=\"T_672bb_row2_col1\" class=\"data row2 col1\" >0.788382</td>\n",
       "      <td id=\"T_672bb_row2_col2\" class=\"data row2 col2\" >-0.017528</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row3\" class=\"row_heading level0 row3\" >3</th>\n",
       "      <td id=\"T_672bb_row3_col0\" class=\"data row3 col0\" >0.007000</td>\n",
       "      <td id=\"T_672bb_row3_col1\" class=\"data row3 col1\" >0.788382</td>\n",
       "      <td id=\"T_672bb_row3_col2\" class=\"data row3 col2\" >-0.017528</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row4\" class=\"row_heading level0 row4\" >4</th>\n",
       "      <td id=\"T_672bb_row4_col0\" class=\"data row4 col0\" >0.009000</td>\n",
       "      <td id=\"T_672bb_row4_col1\" class=\"data row4 col1\" >0.787944</td>\n",
       "      <td id=\"T_672bb_row4_col2\" class=\"data row4 col2\" >-0.011181</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row5\" class=\"row_heading level0 row5\" >5</th>\n",
       "      <td id=\"T_672bb_row5_col0\" class=\"data row5 col0\" >0.011000</td>\n",
       "      <td id=\"T_672bb_row5_col1\" class=\"data row5 col1\" >0.788002</td>\n",
       "      <td id=\"T_672bb_row5_col2\" class=\"data row5 col2\" >-0.018480</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row6\" class=\"row_heading level0 row6\" >6</th>\n",
       "      <td id=\"T_672bb_row6_col0\" class=\"data row6 col0\" >0.013000</td>\n",
       "      <td id=\"T_672bb_row6_col1\" class=\"data row6 col1\" >0.788002</td>\n",
       "      <td id=\"T_672bb_row6_col2\" class=\"data row6 col2\" >-0.018480</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row7\" class=\"row_heading level0 row7\" >7</th>\n",
       "      <td id=\"T_672bb_row7_col0\" class=\"data row7 col0\" >0.015000</td>\n",
       "      <td id=\"T_672bb_row7_col1\" class=\"data row7 col1\" >0.788441</td>\n",
       "      <td id=\"T_672bb_row7_col2\" class=\"data row7 col2\" >-0.024968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row8\" class=\"row_heading level0 row8\" >8</th>\n",
       "      <td id=\"T_672bb_row8_col0\" class=\"data row8 col0\" >0.017000</td>\n",
       "      <td id=\"T_672bb_row8_col1\" class=\"data row8 col1\" >0.788441</td>\n",
       "      <td id=\"T_672bb_row8_col2\" class=\"data row8 col2\" >-0.024968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row9\" class=\"row_heading level0 row9\" >9</th>\n",
       "      <td id=\"T_672bb_row9_col0\" class=\"data row9 col0\" >0.019000</td>\n",
       "      <td id=\"T_672bb_row9_col1\" class=\"data row9 col1\" >0.788441</td>\n",
       "      <td id=\"T_672bb_row9_col2\" class=\"data row9 col2\" >-0.024968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row10\" class=\"row_heading level0 row10\" >10</th>\n",
       "      <td id=\"T_672bb_row10_col0\" class=\"data row10 col0\" >0.021000</td>\n",
       "      <td id=\"T_672bb_row10_col1\" class=\"data row10 col1\" >0.788441</td>\n",
       "      <td id=\"T_672bb_row10_col2\" class=\"data row10 col2\" >-0.024968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row11\" class=\"row_heading level0 row11\" >11</th>\n",
       "      <td id=\"T_672bb_row11_col0\" class=\"data row11 col0\" >0.023000</td>\n",
       "      <td id=\"T_672bb_row11_col1\" class=\"data row11 col1\" >0.788441</td>\n",
       "      <td id=\"T_672bb_row11_col2\" class=\"data row11 col2\" >-0.024968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row12\" class=\"row_heading level0 row12\" >12</th>\n",
       "      <td id=\"T_672bb_row12_col0\" class=\"data row12 col0\" >0.025000</td>\n",
       "      <td id=\"T_672bb_row12_col1\" class=\"data row12 col1\" >0.788441</td>\n",
       "      <td id=\"T_672bb_row12_col2\" class=\"data row12 col2\" >-0.024968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row13\" class=\"row_heading level0 row13\" >13</th>\n",
       "      <td id=\"T_672bb_row13_col0\" class=\"data row13 col0\" >0.027000</td>\n",
       "      <td id=\"T_672bb_row13_col1\" class=\"data row13 col1\" >0.788441</td>\n",
       "      <td id=\"T_672bb_row13_col2\" class=\"data row13 col2\" >-0.024968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row14\" class=\"row_heading level0 row14\" >14</th>\n",
       "      <td id=\"T_672bb_row14_col0\" class=\"data row14 col0\" >0.029000</td>\n",
       "      <td id=\"T_672bb_row14_col1\" class=\"data row14 col1\" >0.788441</td>\n",
       "      <td id=\"T_672bb_row14_col2\" class=\"data row14 col2\" >-0.024968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row15\" class=\"row_heading level0 row15\" >15</th>\n",
       "      <td id=\"T_672bb_row15_col0\" class=\"data row15 col0\" >0.031000</td>\n",
       "      <td id=\"T_672bb_row15_col1\" class=\"data row15 col1\" >0.788441</td>\n",
       "      <td id=\"T_672bb_row15_col2\" class=\"data row15 col2\" >-0.024968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row16\" class=\"row_heading level0 row16\" >16</th>\n",
       "      <td id=\"T_672bb_row16_col0\" class=\"data row16 col0\" >0.033000</td>\n",
       "      <td id=\"T_672bb_row16_col1\" class=\"data row16 col1\" >0.788441</td>\n",
       "      <td id=\"T_672bb_row16_col2\" class=\"data row16 col2\" >-0.024968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row17\" class=\"row_heading level0 row17\" >17</th>\n",
       "      <td id=\"T_672bb_row17_col0\" class=\"data row17 col0\" >0.035000</td>\n",
       "      <td id=\"T_672bb_row17_col1\" class=\"data row17 col1\" >0.788441</td>\n",
       "      <td id=\"T_672bb_row17_col2\" class=\"data row17 col2\" >-0.024968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row18\" class=\"row_heading level0 row18\" >18</th>\n",
       "      <td id=\"T_672bb_row18_col0\" class=\"data row18 col0\" >0.037000</td>\n",
       "      <td id=\"T_672bb_row18_col1\" class=\"data row18 col1\" >0.788441</td>\n",
       "      <td id=\"T_672bb_row18_col2\" class=\"data row18 col2\" >-0.024968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row19\" class=\"row_heading level0 row19\" >19</th>\n",
       "      <td id=\"T_672bb_row19_col0\" class=\"data row19 col0\" >0.039000</td>\n",
       "      <td id=\"T_672bb_row19_col1\" class=\"data row19 col1\" >0.788441</td>\n",
       "      <td id=\"T_672bb_row19_col2\" class=\"data row19 col2\" >-0.024968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row20\" class=\"row_heading level0 row20\" >20</th>\n",
       "      <td id=\"T_672bb_row20_col0\" class=\"data row20 col0\" >0.041000</td>\n",
       "      <td id=\"T_672bb_row20_col1\" class=\"data row20 col1\" >0.788441</td>\n",
       "      <td id=\"T_672bb_row20_col2\" class=\"data row20 col2\" >-0.024968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row21\" class=\"row_heading level0 row21\" >21</th>\n",
       "      <td id=\"T_672bb_row21_col0\" class=\"data row21 col0\" >0.043000</td>\n",
       "      <td id=\"T_672bb_row21_col1\" class=\"data row21 col1\" >0.788441</td>\n",
       "      <td id=\"T_672bb_row21_col2\" class=\"data row21 col2\" >-0.024968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row22\" class=\"row_heading level0 row22\" >22</th>\n",
       "      <td id=\"T_672bb_row22_col0\" class=\"data row22 col0\" >0.045000</td>\n",
       "      <td id=\"T_672bb_row22_col1\" class=\"data row22 col1\" >0.788441</td>\n",
       "      <td id=\"T_672bb_row22_col2\" class=\"data row22 col2\" >-0.024968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row23\" class=\"row_heading level0 row23\" >23</th>\n",
       "      <td id=\"T_672bb_row23_col0\" class=\"data row23 col0\" >0.047000</td>\n",
       "      <td id=\"T_672bb_row23_col1\" class=\"data row23 col1\" >0.788441</td>\n",
       "      <td id=\"T_672bb_row23_col2\" class=\"data row23 col2\" >-0.024968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row24\" class=\"row_heading level0 row24\" >24</th>\n",
       "      <td id=\"T_672bb_row24_col0\" class=\"data row24 col0\" >0.049000</td>\n",
       "      <td id=\"T_672bb_row24_col1\" class=\"data row24 col1\" >0.788441</td>\n",
       "      <td id=\"T_672bb_row24_col2\" class=\"data row24 col2\" >-0.024968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row25\" class=\"row_heading level0 row25\" >25</th>\n",
       "      <td id=\"T_672bb_row25_col0\" class=\"data row25 col0\" >0.051000</td>\n",
       "      <td id=\"T_672bb_row25_col1\" class=\"data row25 col1\" >0.788441</td>\n",
       "      <td id=\"T_672bb_row25_col2\" class=\"data row25 col2\" >-0.024968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row26\" class=\"row_heading level0 row26\" >26</th>\n",
       "      <td id=\"T_672bb_row26_col0\" class=\"data row26 col0\" >0.053000</td>\n",
       "      <td id=\"T_672bb_row26_col1\" class=\"data row26 col1\" >0.788441</td>\n",
       "      <td id=\"T_672bb_row26_col2\" class=\"data row26 col2\" >-0.024968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row27\" class=\"row_heading level0 row27\" >27</th>\n",
       "      <td id=\"T_672bb_row27_col0\" class=\"data row27 col0\" >0.055000</td>\n",
       "      <td id=\"T_672bb_row27_col1\" class=\"data row27 col1\" >0.788441</td>\n",
       "      <td id=\"T_672bb_row27_col2\" class=\"data row27 col2\" >-0.024968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row28\" class=\"row_heading level0 row28\" >28</th>\n",
       "      <td id=\"T_672bb_row28_col0\" class=\"data row28 col0\" >0.057000</td>\n",
       "      <td id=\"T_672bb_row28_col1\" class=\"data row28 col1\" >0.788441</td>\n",
       "      <td id=\"T_672bb_row28_col2\" class=\"data row28 col2\" >-0.024968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row29\" class=\"row_heading level0 row29\" >29</th>\n",
       "      <td id=\"T_672bb_row29_col0\" class=\"data row29 col0\" >0.059000</td>\n",
       "      <td id=\"T_672bb_row29_col1\" class=\"data row29 col1\" >0.788441</td>\n",
       "      <td id=\"T_672bb_row29_col2\" class=\"data row29 col2\" >-0.024968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row30\" class=\"row_heading level0 row30\" >30</th>\n",
       "      <td id=\"T_672bb_row30_col0\" class=\"data row30 col0\" >0.061000</td>\n",
       "      <td id=\"T_672bb_row30_col1\" class=\"data row30 col1\" >0.788441</td>\n",
       "      <td id=\"T_672bb_row30_col2\" class=\"data row30 col2\" >-0.024968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row31\" class=\"row_heading level0 row31\" >31</th>\n",
       "      <td id=\"T_672bb_row31_col0\" class=\"data row31 col0\" >0.063000</td>\n",
       "      <td id=\"T_672bb_row31_col1\" class=\"data row31 col1\" >0.788441</td>\n",
       "      <td id=\"T_672bb_row31_col2\" class=\"data row31 col2\" >-0.024968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row32\" class=\"row_heading level0 row32\" >32</th>\n",
       "      <td id=\"T_672bb_row32_col0\" class=\"data row32 col0\" >0.065000</td>\n",
       "      <td id=\"T_672bb_row32_col1\" class=\"data row32 col1\" >0.788441</td>\n",
       "      <td id=\"T_672bb_row32_col2\" class=\"data row32 col2\" >-0.024968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row33\" class=\"row_heading level0 row33\" >33</th>\n",
       "      <td id=\"T_672bb_row33_col0\" class=\"data row33 col0\" >0.067000</td>\n",
       "      <td id=\"T_672bb_row33_col1\" class=\"data row33 col1\" >0.788441</td>\n",
       "      <td id=\"T_672bb_row33_col2\" class=\"data row33 col2\" >-0.024968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_672bb_level0_row34\" class=\"row_heading level0 row34\" >34</th>\n",
       "      <td id=\"T_672bb_row34_col0\" class=\"data row34 col0\" >0.069000</td>\n",
       "      <td id=\"T_672bb_row34_col1\" class=\"data row34 col1\" >0.788441</td>\n",
       "      <td id=\"T_672bb_row34_col2\" class=\"data row34 col2\" >-0.024968</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n"
      ],
      "text/plain": [
       "<pandas.io.formats.style.Styler at 0x7f32c6752e50>"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def syling(s):\n",
    "  if s['C'] == 0.003:\n",
    "    return ['background-color: orange']*3\n",
    "  else:\n",
    "    return ['background-color: white']*3\n",
    "\n",
    "dfa3.style.apply(syling, axis = 1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "A5yRrsiBfDFb"
   },
   "source": [
    "### German"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 439
    },
    "id": "V8OeOEvzaQJ1",
    "outputId": "3c62173b-a708-4788-d967-4103c1b09bce"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best accuracy score is at C:  0.001 with accuracy score of 0.72\n",
      "Best Equality of Opportunity score is at C: 0.027000000000000003 with score -0.000684931506849451\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#reweighted cross fold on german dataset with refined c\n",
    "\n",
    "dfg3 = cv_rw_fairness(X_train_g, y_train_g, g_train, a)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "w0FTHxfkUJuI"
   },
   "source": [
    "**German Dataset:** (Re-weighing) The hyperparameter for *Equality of Opportunity* is at **C: 0.027** with a score of **-0.0006849**\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "id": "HgF3ZuGgSiic",
    "outputId": "353804c5-405c-48c2-9674-b851d0209b92"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<style type=\"text/css\">\n",
       "#T_f30ce_row0_col0, #T_f30ce_row0_col1, #T_f30ce_row0_col2, #T_f30ce_row1_col0, #T_f30ce_row1_col1, #T_f30ce_row1_col2, #T_f30ce_row2_col0, #T_f30ce_row2_col1, #T_f30ce_row2_col2, #T_f30ce_row3_col0, #T_f30ce_row3_col1, #T_f30ce_row3_col2, #T_f30ce_row4_col0, #T_f30ce_row4_col1, #T_f30ce_row4_col2, #T_f30ce_row5_col0, #T_f30ce_row5_col1, #T_f30ce_row5_col2, #T_f30ce_row6_col0, #T_f30ce_row6_col1, #T_f30ce_row6_col2, #T_f30ce_row7_col0, #T_f30ce_row7_col1, #T_f30ce_row7_col2, #T_f30ce_row8_col0, #T_f30ce_row8_col1, #T_f30ce_row8_col2, #T_f30ce_row9_col0, #T_f30ce_row9_col1, #T_f30ce_row9_col2, #T_f30ce_row10_col0, #T_f30ce_row10_col1, #T_f30ce_row10_col2, #T_f30ce_row11_col0, #T_f30ce_row11_col1, #T_f30ce_row11_col2, #T_f30ce_row12_col0, #T_f30ce_row12_col1, #T_f30ce_row12_col2, #T_f30ce_row14_col0, #T_f30ce_row14_col1, #T_f30ce_row14_col2, #T_f30ce_row15_col0, #T_f30ce_row15_col1, #T_f30ce_row15_col2, #T_f30ce_row16_col0, #T_f30ce_row16_col1, #T_f30ce_row16_col2, #T_f30ce_row17_col0, #T_f30ce_row17_col1, #T_f30ce_row17_col2, #T_f30ce_row18_col0, #T_f30ce_row18_col1, #T_f30ce_row18_col2, #T_f30ce_row19_col0, #T_f30ce_row19_col1, #T_f30ce_row19_col2, #T_f30ce_row20_col0, #T_f30ce_row20_col1, #T_f30ce_row20_col2, #T_f30ce_row21_col0, #T_f30ce_row21_col1, #T_f30ce_row21_col2, #T_f30ce_row22_col0, #T_f30ce_row22_col1, #T_f30ce_row22_col2, #T_f30ce_row23_col0, #T_f30ce_row23_col1, #T_f30ce_row23_col2, #T_f30ce_row24_col0, #T_f30ce_row24_col1, #T_f30ce_row24_col2, #T_f30ce_row25_col0, #T_f30ce_row25_col1, #T_f30ce_row25_col2, #T_f30ce_row26_col0, #T_f30ce_row26_col1, #T_f30ce_row26_col2, #T_f30ce_row27_col0, #T_f30ce_row27_col1, #T_f30ce_row27_col2, #T_f30ce_row28_col0, #T_f30ce_row28_col1, #T_f30ce_row28_col2, #T_f30ce_row29_col0, #T_f30ce_row29_col1, #T_f30ce_row29_col2, #T_f30ce_row30_col0, #T_f30ce_row30_col1, #T_f30ce_row30_col2, #T_f30ce_row31_col0, #T_f30ce_row31_col1, #T_f30ce_row31_col2, #T_f30ce_row32_col0, #T_f30ce_row32_col1, #T_f30ce_row32_col2, #T_f30ce_row33_col0, #T_f30ce_row33_col1, #T_f30ce_row33_col2, #T_f30ce_row34_col0, #T_f30ce_row34_col1, #T_f30ce_row34_col2 {\n",
       "  background-color: white;\n",
       "}\n",
       "#T_f30ce_row13_col0, #T_f30ce_row13_col1, #T_f30ce_row13_col2 {\n",
       "  background-color: orange;\n",
       "}\n",
       "</style>\n",
       "<table id=\"T_f30ce_\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th class=\"blank level0\" >&nbsp;</th>\n",
       "      <th class=\"col_heading level0 col0\" >C</th>\n",
       "      <th class=\"col_heading level0 col1\" >Accuracy</th>\n",
       "      <th class=\"col_heading level0 col2\" >Equality of Opportunity</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row0\" class=\"row_heading level0 row0\" >0</th>\n",
       "      <td id=\"T_f30ce_row0_col0\" class=\"data row0 col0\" >0.001000</td>\n",
       "      <td id=\"T_f30ce_row0_col1\" class=\"data row0 col1\" >0.720000</td>\n",
       "      <td id=\"T_f30ce_row0_col2\" class=\"data row0 col2\" >-0.039237</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row1\" class=\"row_heading level0 row1\" >1</th>\n",
       "      <td id=\"T_f30ce_row1_col0\" class=\"data row1 col0\" >0.003000</td>\n",
       "      <td id=\"T_f30ce_row1_col1\" class=\"data row1 col1\" >0.718571</td>\n",
       "      <td id=\"T_f30ce_row1_col2\" class=\"data row1 col2\" >-0.032094</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row2\" class=\"row_heading level0 row2\" >2</th>\n",
       "      <td id=\"T_f30ce_row2_col0\" class=\"data row2 col0\" >0.005000</td>\n",
       "      <td id=\"T_f30ce_row2_col1\" class=\"data row2 col1\" >0.711429</td>\n",
       "      <td id=\"T_f30ce_row2_col2\" class=\"data row2 col2\" >-0.007339</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row3\" class=\"row_heading level0 row3\" >3</th>\n",
       "      <td id=\"T_f30ce_row3_col0\" class=\"data row3 col0\" >0.007000</td>\n",
       "      <td id=\"T_f30ce_row3_col1\" class=\"data row3 col1\" >0.710000</td>\n",
       "      <td id=\"T_f30ce_row3_col2\" class=\"data row3 col2\" >-0.018297</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row4\" class=\"row_heading level0 row4\" >4</th>\n",
       "      <td id=\"T_f30ce_row4_col0\" class=\"data row4 col0\" >0.009000</td>\n",
       "      <td id=\"T_f30ce_row4_col1\" class=\"data row4 col1\" >0.705714</td>\n",
       "      <td id=\"T_f30ce_row4_col2\" class=\"data row4 col2\" >0.003131</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row5\" class=\"row_heading level0 row5\" >5</th>\n",
       "      <td id=\"T_f30ce_row5_col0\" class=\"data row5 col0\" >0.011000</td>\n",
       "      <td id=\"T_f30ce_row5_col1\" class=\"data row5 col1\" >0.710000</td>\n",
       "      <td id=\"T_f30ce_row5_col2\" class=\"data row5 col2\" >-0.007828</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row6\" class=\"row_heading level0 row6\" >6</th>\n",
       "      <td id=\"T_f30ce_row6_col0\" class=\"data row6 col0\" >0.013000</td>\n",
       "      <td id=\"T_f30ce_row6_col1\" class=\"data row6 col1\" >0.710000</td>\n",
       "      <td id=\"T_f30ce_row6_col2\" class=\"data row6 col2\" >-0.007828</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row7\" class=\"row_heading level0 row7\" >7</th>\n",
       "      <td id=\"T_f30ce_row7_col0\" class=\"data row7 col0\" >0.015000</td>\n",
       "      <td id=\"T_f30ce_row7_col1\" class=\"data row7 col1\" >0.710000</td>\n",
       "      <td id=\"T_f30ce_row7_col2\" class=\"data row7 col2\" >-0.007828</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row8\" class=\"row_heading level0 row8\" >8</th>\n",
       "      <td id=\"T_f30ce_row8_col0\" class=\"data row8 col0\" >0.017000</td>\n",
       "      <td id=\"T_f30ce_row8_col1\" class=\"data row8 col1\" >0.710000</td>\n",
       "      <td id=\"T_f30ce_row8_col2\" class=\"data row8 col2\" >-0.007828</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row9\" class=\"row_heading level0 row9\" >9</th>\n",
       "      <td id=\"T_f30ce_row9_col0\" class=\"data row9 col0\" >0.019000</td>\n",
       "      <td id=\"T_f30ce_row9_col1\" class=\"data row9 col1\" >0.710000</td>\n",
       "      <td id=\"T_f30ce_row9_col2\" class=\"data row9 col2\" >-0.007828</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row10\" class=\"row_heading level0 row10\" >10</th>\n",
       "      <td id=\"T_f30ce_row10_col0\" class=\"data row10 col0\" >0.021000</td>\n",
       "      <td id=\"T_f30ce_row10_col1\" class=\"data row10 col1\" >0.710000</td>\n",
       "      <td id=\"T_f30ce_row10_col2\" class=\"data row10 col2\" >-0.007828</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row11\" class=\"row_heading level0 row11\" >11</th>\n",
       "      <td id=\"T_f30ce_row11_col0\" class=\"data row11 col0\" >0.023000</td>\n",
       "      <td id=\"T_f30ce_row11_col1\" class=\"data row11 col1\" >0.707143</td>\n",
       "      <td id=\"T_f30ce_row11_col2\" class=\"data row11 col2\" >-0.007828</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row12\" class=\"row_heading level0 row12\" >12</th>\n",
       "      <td id=\"T_f30ce_row12_col0\" class=\"data row12 col0\" >0.025000</td>\n",
       "      <td id=\"T_f30ce_row12_col1\" class=\"data row12 col1\" >0.707143</td>\n",
       "      <td id=\"T_f30ce_row12_col2\" class=\"data row12 col2\" >-0.007828</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row13\" class=\"row_heading level0 row13\" >13</th>\n",
       "      <td id=\"T_f30ce_row13_col0\" class=\"data row13 col0\" >0.027000</td>\n",
       "      <td id=\"T_f30ce_row13_col1\" class=\"data row13 col1\" >0.704286</td>\n",
       "      <td id=\"T_f30ce_row13_col2\" class=\"data row13 col2\" >-0.000685</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row14\" class=\"row_heading level0 row14\" >14</th>\n",
       "      <td id=\"T_f30ce_row14_col0\" class=\"data row14 col0\" >0.029000</td>\n",
       "      <td id=\"T_f30ce_row14_col1\" class=\"data row14 col1\" >0.704286</td>\n",
       "      <td id=\"T_f30ce_row14_col2\" class=\"data row14 col2\" >-0.000685</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row15\" class=\"row_heading level0 row15\" >15</th>\n",
       "      <td id=\"T_f30ce_row15_col0\" class=\"data row15 col0\" >0.031000</td>\n",
       "      <td id=\"T_f30ce_row15_col1\" class=\"data row15 col1\" >0.704286</td>\n",
       "      <td id=\"T_f30ce_row15_col2\" class=\"data row15 col2\" >-0.000685</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row16\" class=\"row_heading level0 row16\" >16</th>\n",
       "      <td id=\"T_f30ce_row16_col0\" class=\"data row16 col0\" >0.033000</td>\n",
       "      <td id=\"T_f30ce_row16_col1\" class=\"data row16 col1\" >0.705714</td>\n",
       "      <td id=\"T_f30ce_row16_col2\" class=\"data row16 col2\" >-0.000685</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row17\" class=\"row_heading level0 row17\" >17</th>\n",
       "      <td id=\"T_f30ce_row17_col0\" class=\"data row17 col0\" >0.035000</td>\n",
       "      <td id=\"T_f30ce_row17_col1\" class=\"data row17 col1\" >0.705714</td>\n",
       "      <td id=\"T_f30ce_row17_col2\" class=\"data row17 col2\" >-0.000685</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row18\" class=\"row_heading level0 row18\" >18</th>\n",
       "      <td id=\"T_f30ce_row18_col0\" class=\"data row18 col0\" >0.037000</td>\n",
       "      <td id=\"T_f30ce_row18_col1\" class=\"data row18 col1\" >0.708571</td>\n",
       "      <td id=\"T_f30ce_row18_col2\" class=\"data row18 col2\" >0.013601</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row19\" class=\"row_heading level0 row19\" >19</th>\n",
       "      <td id=\"T_f30ce_row19_col0\" class=\"data row19 col0\" >0.039000</td>\n",
       "      <td id=\"T_f30ce_row19_col1\" class=\"data row19 col1\" >0.708571</td>\n",
       "      <td id=\"T_f30ce_row19_col2\" class=\"data row19 col2\" >0.013601</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row20\" class=\"row_heading level0 row20\" >20</th>\n",
       "      <td id=\"T_f30ce_row20_col0\" class=\"data row20 col0\" >0.041000</td>\n",
       "      <td id=\"T_f30ce_row20_col1\" class=\"data row20 col1\" >0.708571</td>\n",
       "      <td id=\"T_f30ce_row20_col2\" class=\"data row20 col2\" >0.013601</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row21\" class=\"row_heading level0 row21\" >21</th>\n",
       "      <td id=\"T_f30ce_row21_col0\" class=\"data row21 col0\" >0.043000</td>\n",
       "      <td id=\"T_f30ce_row21_col1\" class=\"data row21 col1\" >0.708571</td>\n",
       "      <td id=\"T_f30ce_row21_col2\" class=\"data row21 col2\" >0.013601</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row22\" class=\"row_heading level0 row22\" >22</th>\n",
       "      <td id=\"T_f30ce_row22_col0\" class=\"data row22 col0\" >0.045000</td>\n",
       "      <td id=\"T_f30ce_row22_col1\" class=\"data row22 col1\" >0.708571</td>\n",
       "      <td id=\"T_f30ce_row22_col2\" class=\"data row22 col2\" >0.013601</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row23\" class=\"row_heading level0 row23\" >23</th>\n",
       "      <td id=\"T_f30ce_row23_col0\" class=\"data row23 col0\" >0.047000</td>\n",
       "      <td id=\"T_f30ce_row23_col1\" class=\"data row23 col1\" >0.708571</td>\n",
       "      <td id=\"T_f30ce_row23_col2\" class=\"data row23 col2\" >0.013601</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row24\" class=\"row_heading level0 row24\" >24</th>\n",
       "      <td id=\"T_f30ce_row24_col0\" class=\"data row24 col0\" >0.049000</td>\n",
       "      <td id=\"T_f30ce_row24_col1\" class=\"data row24 col1\" >0.708571</td>\n",
       "      <td id=\"T_f30ce_row24_col2\" class=\"data row24 col2\" >0.013601</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row25\" class=\"row_heading level0 row25\" >25</th>\n",
       "      <td id=\"T_f30ce_row25_col0\" class=\"data row25 col0\" >0.051000</td>\n",
       "      <td id=\"T_f30ce_row25_col1\" class=\"data row25 col1\" >0.708571</td>\n",
       "      <td id=\"T_f30ce_row25_col2\" class=\"data row25 col2\" >0.013601</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row26\" class=\"row_heading level0 row26\" >26</th>\n",
       "      <td id=\"T_f30ce_row26_col0\" class=\"data row26 col0\" >0.053000</td>\n",
       "      <td id=\"T_f30ce_row26_col1\" class=\"data row26 col1\" >0.708571</td>\n",
       "      <td id=\"T_f30ce_row26_col2\" class=\"data row26 col2\" >0.013601</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row27\" class=\"row_heading level0 row27\" >27</th>\n",
       "      <td id=\"T_f30ce_row27_col0\" class=\"data row27 col0\" >0.055000</td>\n",
       "      <td id=\"T_f30ce_row27_col1\" class=\"data row27 col1\" >0.710000</td>\n",
       "      <td id=\"T_f30ce_row27_col2\" class=\"data row27 col2\" >0.013601</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row28\" class=\"row_heading level0 row28\" >28</th>\n",
       "      <td id=\"T_f30ce_row28_col0\" class=\"data row28 col0\" >0.057000</td>\n",
       "      <td id=\"T_f30ce_row28_col1\" class=\"data row28 col1\" >0.710000</td>\n",
       "      <td id=\"T_f30ce_row28_col2\" class=\"data row28 col2\" >0.013601</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row29\" class=\"row_heading level0 row29\" >29</th>\n",
       "      <td id=\"T_f30ce_row29_col0\" class=\"data row29 col0\" >0.059000</td>\n",
       "      <td id=\"T_f30ce_row29_col1\" class=\"data row29 col1\" >0.710000</td>\n",
       "      <td id=\"T_f30ce_row29_col2\" class=\"data row29 col2\" >0.013601</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row30\" class=\"row_heading level0 row30\" >30</th>\n",
       "      <td id=\"T_f30ce_row30_col0\" class=\"data row30 col0\" >0.061000</td>\n",
       "      <td id=\"T_f30ce_row30_col1\" class=\"data row30 col1\" >0.710000</td>\n",
       "      <td id=\"T_f30ce_row30_col2\" class=\"data row30 col2\" >0.013601</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row31\" class=\"row_heading level0 row31\" >31</th>\n",
       "      <td id=\"T_f30ce_row31_col0\" class=\"data row31 col0\" >0.063000</td>\n",
       "      <td id=\"T_f30ce_row31_col1\" class=\"data row31 col1\" >0.710000</td>\n",
       "      <td id=\"T_f30ce_row31_col2\" class=\"data row31 col2\" >0.013601</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row32\" class=\"row_heading level0 row32\" >32</th>\n",
       "      <td id=\"T_f30ce_row32_col0\" class=\"data row32 col0\" >0.065000</td>\n",
       "      <td id=\"T_f30ce_row32_col1\" class=\"data row32 col1\" >0.710000</td>\n",
       "      <td id=\"T_f30ce_row32_col2\" class=\"data row32 col2\" >0.013601</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row33\" class=\"row_heading level0 row33\" >33</th>\n",
       "      <td id=\"T_f30ce_row33_col0\" class=\"data row33 col0\" >0.067000</td>\n",
       "      <td id=\"T_f30ce_row33_col1\" class=\"data row33 col1\" >0.710000</td>\n",
       "      <td id=\"T_f30ce_row33_col2\" class=\"data row33 col2\" >0.013601</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th id=\"T_f30ce_level0_row34\" class=\"row_heading level0 row34\" >34</th>\n",
       "      <td id=\"T_f30ce_row34_col0\" class=\"data row34 col0\" >0.069000</td>\n",
       "      <td id=\"T_f30ce_row34_col1\" class=\"data row34 col1\" >0.710000</td>\n",
       "      <td id=\"T_f30ce_row34_col2\" class=\"data row34 col2\" >0.013601</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n"
      ],
      "text/plain": [
       "<pandas.io.formats.style.Styler at 0x7f32c66a1990>"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def syling(s):\n",
    "  if s['C'] == 0.027000000000000003:\n",
    "    return ['background-color: orange']*3\n",
    "  else:\n",
    "    return ['background-color: white']*3\n",
    "\n",
    "dfg3.style.apply(syling, axis = 1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "WjO9ArX1ftjK"
   },
   "source": [
    "## Compute and Report the Final Accuracy: Re-weighing\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "JooES0AqQ9lF"
   },
   "source": [
    "### Task 2: Table of Results from Training Data\n",
    "\n",
    "| Dataset | C Value for Highest Accuracy (3) | C Value for Best Fairness Metrics (4) |\n",
    "| --- | :---: | ---: |\n",
    "| **Adult** | 0.001 | 0.003 |\n",
    "| **German** | 0.12 | 0.027 |"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "sU850GtIgZ8w"
   },
   "source": [
    "### Functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "id": "LT9kLFPQhmIe"
   },
   "outputs": [],
   "source": [
    "#reweighing final scores\n",
    "\n",
    "def rw_finalscores(X_train, y_train, X_test, y_test,train, test, c):\n",
    "  equ_opp_list = []\n",
    "  acc = []\n",
    "  disparate = []\n",
    "  #mitgating the bias\n",
    "  #assigning the new weights to the unprivileged and privileged groups, transform the dataset with reweighing\n",
    "  RW = Reweighing(unprivileged_groups=unprivileged_groups,\n",
    "                privileged_groups=privileged_groups)\n",
    "  #obtaining a set of weights for the training set\n",
    "  train = RW.fit_transform(train)\n",
    "  clf = LogisticRegression(solver='liblinear', C=c, random_state=1)\n",
    "  clf.fit(X_train,y_train,sample_weight=train.instance_weights)\n",
    "  y_pred = clf.predict(X_test)\n",
    "  test_pred = test.copy()\n",
    "  test_pred.labels = y_pred\n",
    "  metric = ClassificationMetric(test, test_pred, unprivileged_groups=unprivileged_groups, privileged_groups=privileged_groups)\n",
    "  \n",
    "  #storing the results\n",
    "  acc.append(sum(y_pred==y_test)/len(y_test))\n",
    "  equ_opp_list.append(metric.equal_opportunity_difference()) #would like it to be close to 0\n",
    "  disparate.append(metric.disparate_impact()) #value close to 0 indicates strong bias, would like to be close to 1\n",
    "\n",
    "  #printing the results\n",
    "  print('After reweighing, the final accuracy score on the test data with hyperparameter C:', c, 'is', clf.score(X_test, y_test))\n",
    "  print('The Equality of Opportunity score at C:', c, 'is', equ_opp_list)\n",
    "  print('The disparate impact score is', disparate)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "fUrg1AT_iEHv"
   },
   "source": [
    "### Adult"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "i7dWK1ErSBiO"
   },
   "source": [
    "<font color='purple'>**(3)**</font> The hyperparameter for the highest accuracy on training data is at **C: 0.001**\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "jN1W72q9iGbG",
    "outputId": "f42672f7-f980-429d-95b6-08bcf3b7af9b"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "After reweighing, the final accuracy score on the test data with hyperparameter C: 0.001 is 0.7919197433972566\n",
      "The Equality of Opportunity score at C: 0.001 is [0.03607506672931393]\n",
      "The disparate impact score is [0.7025889736823336]\n"
     ]
    }
   ],
   "source": [
    "#adult final scores at hyperparameter 0.001\n",
    "\n",
    "rw_finalscores(X_train_a, y_train_a, X_test_a, y_test_a,a_train, a_test, 0.001)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "2_QduZpYVi9N"
   },
   "source": [
    "> After reweighing, the <font color='purple'>**final *accuracy* score**</font> in testing data at hyperparameter C: 0.001 is <font color='purple'>**0.79192**</font>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "ur1eHuD8SrCd"
   },
   "source": [
    "<font color='darkorange'>**(4)**</font> The hyperparameter for the best Equality of Opportunity score on training data is at **C: 0.003**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "GacEA20aS5X4",
    "outputId": "4253cbd5-72ee-4666-c9da-fc7cc638727a"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "After reweighing, the final accuracy score on the test data with hyperparameter C: 0.003 is 0.7904183443663414\n",
      "The Equality of Opportunity score at C: 0.003 is [0.050140706381021916]\n",
      "The disparate impact score is [0.7250312798999043]\n"
     ]
    }
   ],
   "source": [
    "#adult final scores at hyperparameter 0.003\n",
    "\n",
    "rw_finalscores(X_train_a, y_train_a, X_test_a, y_test_a,a_train, a_test, 0.003)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "UoLamsl-V64w"
   },
   "source": [
    "> After reweighing, the <font color='orange'>**final *Equality of Opportunity* score**</font> on testing data at hyperparameter C: 0.003 is <font color='orange'>**0.05014**</font>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "m6PzBie4idbO"
   },
   "source": [
    "### German"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "VMTIjL_TTP9Z"
   },
   "source": [
    "<font color='purple'>**(3)**</font> The hyperparameter for the highest accuracy on training data is at **C: 0.12**\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "xew1qWa9TTEd",
    "outputId": "5049f1f9-6791-4d74-d3c6-4879d26e1a6b"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "After reweighing, the final accuracy score on the test data with hyperparameter C: 0.12 is 0.6366666666666667\n",
      "The Equality of Opportunity score at C: 0.12 is [-0.002936138977244984]\n",
      "The disparate impact score is [0.9815705128205129]\n"
     ]
    }
   ],
   "source": [
    "#german final scores at hyperparameter 0.012\n",
    "\n",
    "rw_finalscores(X_train_g, y_train_g, X_test_g, y_test_g,g_train, g_test, 0.12)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "uzjZCnYxVul8"
   },
   "source": [
    "> After reweighing, the <font color='purple'>**final *accuracy* score**</font> in testing data at hyperparameter C: 0.12 is <font color='purple'>**0.6566**</font>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "cqYBcWIW-FA1"
   },
   "source": [
    "<font color='darkorange'>**(4)**</font> The hyperparameter for the best Equality of Opportunity score on training data is at **C: 0.027**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "2gw68wJ9ie1a",
    "outputId": "d45a5eb7-5e1c-4849-c94a-697ddd7ae0e4"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "After reweighing, the final accuracy score on the test data with hyperparameter C: 0.027 is 0.6366666666666667\n",
      "The Equality of Opportunity score at C: 0.027 is [-0.002936138977244984]\n",
      "The disparate impact score is [0.9815705128205129]\n"
     ]
    }
   ],
   "source": [
    "#german final scores at hyperparameter 0.027\n",
    "\n",
    "rw_finalscores(X_train_g, y_train_g, X_test_g, y_test_g,g_train, g_test, 0.027)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "_ugc1OiqWFsi"
   },
   "source": [
    "> After reweighing, the <font color='orange'>**final *Equality of Opportunity* score**</font> on testing data at hyperparameter C: 0.027 is <font color='orange'>**-0.0029**</font>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "TxzClL920H7U"
   },
   "source": [
    "---"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "SOYQG-zqDA5b"
   },
   "source": [
    "# **Task 3a (Extra)**\n",
    "\n",
    "Based on your observations, suggest a model selection strategy (criterion)\n",
    "that accounts for both, accuracy and fairness. Compare the standard machine learning model\n",
    "versus the fairness-based machine learning model using the proposed criterion - what is the effect\n",
    "(if any)?\n",
    "Specifically, using the same 5-fold cross validation as before, select (5) the standard machine\n",
    "learning model (studied in Task 1), and (6) the fairness-based model (studied in Task 2) based\n",
    "on the proposed accuracy+fairness criterion. After model selection, compute and report final\n",
    "accuracy and fairness for both models (5) and (6) on the test set."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "q5xTwi9vNPxQ"
   },
   "source": [
    "What will I be looking at?\n",
    "\n",
    "For part (a) I shall be observing the combination of accuracy and fairness and comparing this with a standard machine learning model and a fairness based machine learning model. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "W1jcZ6c76hSQ"
   },
   "source": [
    "## (5a) Standard Machine Learning Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "GzAlbTtUVwjc"
   },
   "source": [
    "### Functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "id": "piXYuKXbFmjY"
   },
   "outputs": [],
   "source": [
    "#need to find a sweetspot between the fairness and accuracy\n",
    "#we want high accuracy and low equal opp \n",
    "#we would then want high 1-(low equal opp)\n",
    "#add (accuracy and high 1-(low equal opp)) - 1 would be the best combined score\n",
    "\n",
    "#looking into the f1 score \n",
    "\n",
    "def fairaccuracyf1(X_train, y_train, train, cvals):\n",
    "  #empty lists for our scores\n",
    "  acc_scores = []\n",
    "  fairness = []\n",
    "  equ_opp = []\n",
    "  avg_acc_scores = []\n",
    "  f1score = []\n",
    "\n",
    "#looping through each C value\n",
    "  for i in cvals:\n",
    "    #our model\n",
    "    clf = LogisticRegression(solver='liblinear', C=i, random_state=1)\n",
    "    #accuracy score\n",
    "    score = cross_val_score(clf, X_train, y_train, cv = 5)\n",
    "    #prediction labels for fairness\n",
    "    fair_pred = cross_val_predict(clf, X_train, y_train, cv = 5)\n",
    "    test_pred = train.copy()\n",
    "    test_pred.labels = fair_pred\n",
    "\n",
    "    #defining our metric\n",
    "    metric = ClassificationMetric(train, test_pred, unprivileged_groups=unprivileged_groups, privileged_groups=privileged_groups)\n",
    "\n",
    "    #recording the scores\n",
    "    equ_opp.append(metric.equal_opportunity_difference())\n",
    "    f1score.append(f1_score(y_train, fair_pred, average = 'macro'))\n",
    "    avg_acc = np.mean(score)\n",
    "    avg_acc_scores.append(np.mean(score))\n",
    "\n",
    "  #taking the absolute value for the Equality of Opportunity scores\n",
    "  absol_equ_opp = []\n",
    "  for value in equ_opp:\n",
    "    absolute = abs(value)\n",
    "    absol_equ_opp.append(absolute)\n",
    "\n",
    "  #combining the values of accuracy and Equality of Opportunity\n",
    "  combined = [i + (1-j) for (i, j) in zip(avg_acc_scores , absol_equ_opp)]\n",
    "\n",
    "\n",
    "  #re-scaling the combined value to be out of 1, closer to 1 is better\n",
    "  rescaledcombined = []\n",
    "  for value in combined:\n",
    "    rescaled = value - 1\n",
    "    rescaledcombined.append(rescaled)\n",
    "  \n",
    "  #the best scores\n",
    "  best_c = equ_opp.index(min(equ_opp, key= abs))\n",
    "  best_acc_c = avg_acc_scores.index(max(avg_acc_scores)) # we want this to be close to 1\n",
    "  best_combined = rescaledcombined.index(max(rescaledcombined)) # we want this to be close to 1\n",
    "  best_f1 = f1score.index(max(f1score)) #we want f1 score to be 1\n",
    "\n",
    "  print('Best accuracy score is at C: ', cvals[best_acc_c], 'with accuracy score of', max(avg_acc_scores))\n",
    "  print('Best Equality of Opportunity score is at C: ', cvals[best_c], 'with an equality of opportunity score of', min(equ_opp, key= abs))\n",
    "  print('Best f1 score is at C: ', cvals[best_f1], 'with a score of', max(f1score))\n",
    "  print('Best combined score is at C: ', cvals[best_combined], 'with a score of', max(rescaledcombined))\n",
    "  \n",
    "\n",
    "  #plotting the results\n",
    "  xplt = cvals\n",
    "  plt.figure(figsize = (10,6))\n",
    "  plt.scatter(xplt, rescaledcombined, color = 'darkviolet', label = 'Combined: Average and Equality of Opportunity')\n",
    "  plt.scatter(xplt, f1score, color = 'red', label = 'F1 Score')\n",
    "  plt.legend(loc='upper center', bbox_to_anchor=(0.5, -0.10),\n",
    "          fancybox=True, shadow=True)\n",
    "\n",
    "  font1 = {'family':'serif','color':'darkviolet','size':13}\n",
    "  font2 = {'family':'serif','color':'darkviolet','size':11}\n",
    "  plt.title('Sweet Spot', fontdict = font1, color = 'darkviolet')\n",
    "  plt.xlabel('C Value', fontdict = font2)\n",
    "  plt.ylabel('Score', fontdict = font2)\n",
    "  plt.show()\n",
    "  \n",
    "\n",
    "  #return(df) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "ffXCK43u6JFT"
   },
   "source": [
    "### Adult"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 513
    },
    "id": "24YtVZhlCVMC",
    "outputId": "2e386ef9-f0be-4d66-f212-38a2c72173c7"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best accuracy score is at C:  0.05 with accuracy score of 0.8043816906811315\n",
      "Best Equality of Opportunity score is at C:  0.005 with an equality of opportunity score of -0.45715522215810733\n",
      "Best f1 score is at C:  0.05 with a score of 0.6829809555455333\n",
      "Best combined score is at C:  0.005 with a score of 0.347167971886581\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#plotting results for a broud c\n",
    "\n",
    "fairaccuracyf1(X_train_a, y_train_a, a_train, c)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "-7qH1C1NCnRW"
   },
   "source": [
    "Exploring around 0.05"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 513
    },
    "id": "Sw-mtlTvGB4z",
    "outputId": "58f29c39-6018-44d5-fbb3-c33c84abadd7"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best accuracy score is at C:  0.021 with accuracy score of 0.8043816906811315\n",
      "Best Equality of Opportunity score is at C:  0.001 with an equality of opportunity score of -0.4464800923254472\n",
      "Best f1 score is at C:  0.021 with a score of 0.6829809555455333\n",
      "Best combined score is at C:  0.001 with a score of 0.3574336252641368\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#plotting results for a more refined c\n",
    "\n",
    "fairaccuracyf1(X_train_a, y_train_a, a_train, np.arange(0.001, 0.08, 0.005))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "ojet9Pt6C4WF"
   },
   "source": [
    "Exploring around 0.021"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 513
    },
    "id": "HovmFcpE-b8C",
    "outputId": "0d31a2c3-f0cb-4c79-fe1b-f3e38d35ddfd"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best accuracy score is at C:  0.016999999999999994 with accuracy score of 0.8043816906811315\n",
      "Best Equality of Opportunity score is at C:  0.01 with an equality of opportunity score of -0.45715522215810733\n",
      "Best f1 score is at C:  0.016999999999999994 with a score of 0.6829809555455333\n",
      "Best combined score is at C:  0.01 with a score of 0.347167971886581\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#plotting results for a more refined c\n",
    "\n",
    "fairaccuracyf1(X_train_a, y_train_a, a_train, np.arange(0.01, 0.03, 0.001))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "mNsGlREi7Pjl"
   },
   "source": [
    "### German"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 513
    },
    "id": "iMpu3HkNEBCF",
    "outputId": "156223f6-75c2-467c-cd44-cb36dc75dc71"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best accuracy score is at C:  0.01 with accuracy score of 0.7185714285714286\n",
      "Best Equality of Opportunity score is at C:  0.05 with an equality of opportunity score of -0.050097847358121395\n",
      "Best f1 score is at C:  0.005 with a score of 0.5569887959190813\n",
      "Best combined score is at C:  0.05 with a score of 0.6656164383561642\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#plotting results for a broud c\n",
    "\n",
    "fairaccuracyf1(X_train_g, y_train_g, g_train, c)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 513
    },
    "id": "-A-PffasDmtf",
    "outputId": "edc70b2e-433b-4d66-8542-2d4f096669dc"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best accuracy score is at C:  0.009000000000000001 with accuracy score of 0.7185714285714286\n",
      "Best Equality of Opportunity score is at C:  0.05 with an equality of opportunity score of -0.050097847358121395\n",
      "Best f1 score is at C:  0.001 with a score of 0.5719406041986688\n",
      "Best combined score is at C:  0.05 with a score of 0.6656164383561642\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#plotting results for a more refined c around 0.005\n",
    "\n",
    "fairaccuracyf1(X_train_g, y_train_g, g_train,  np.arange(0.001, 0.08, 0.001))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 513
    },
    "id": "Hyuso29WD4ky",
    "outputId": "1f7f4e23-759d-4976-8d99-32d94b9ee3a1"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best accuracy score is at C:  0.008199999999999999 with accuracy score of 0.7185714285714286\n",
      "Best Equality of Opportunity score is at C:  0.0028 with an equality of opportunity score of -0.05352250489236787\n",
      "Best f1 score is at C:  0.001 with a score of 0.5719406041986688\n",
      "Best combined score is at C:  0.0028 with a score of 0.6636203522504893\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#plotting results for a more refined c around 0.005\n",
    "\n",
    "fairaccuracyf1(X_train_g, y_train_g, g_train,  np.arange(0.001, 0.03, 0.0009))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 513
    },
    "id": "mt9YLpooehzO",
    "outputId": "cf3072ac-4499-4baa-85a6-907616405e19"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best accuracy score is at C:  0.011 with accuracy score of 0.7185714285714286\n",
      "Best Equality of Opportunity score is at C:  0.051000000000000004 with an equality of opportunity score of -0.050097847358121395\n",
      "Best f1 score is at C:  0.001 with a score of 0.5719406041986688\n",
      "Best combined score is at C:  0.051000000000000004 with a score of 0.6656164383561642\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#plotting results for a more refined c around 0.0028\n",
    "\n",
    "fairaccuracyf1(X_train_g, y_train_g, g_train,  np.arange(0.001, 0.08, 0.005))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "iUGqKHPnDE69"
   },
   "source": [
    "## (6a) Fairness-based Machine Learning Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "U3wADwyW7S9p"
   },
   "source": [
    "### Functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {
    "id": "jjix1AKWvvVi"
   },
   "outputs": [],
   "source": [
    "#need to find a sweetspot between the fairness and accuracy\n",
    "#we want high accuracy and low equal opp \n",
    "#we would then want high 1-(low equal opp)\n",
    "#add (accuracy and high 1-(low equal opp)) - 1 would be the best combined score\n",
    "\n",
    "#looking into the f1 score \n",
    "\n",
    "def rw_fairaccuracyf1(X_train, y_train, train, cvals):\n",
    "  #empty lists for our scores\n",
    "  acc_scores = []\n",
    "  fairness = []\n",
    "  equ_opp = []\n",
    "  avg_acc_scores = []\n",
    "  f1score = []\n",
    "  #mitgating the bias\n",
    "  #assigning the new weights to the unprivileged and privileged groups, transform the dataset with reweighing\n",
    "  RW = Reweighing(unprivileged_groups=unprivileged_groups,\n",
    "                privileged_groups=privileged_groups)\n",
    "  #obtaining a set of weights for the training set\n",
    "  train = RW.fit_transform(train)\n",
    "\n",
    "\n",
    "#looping through each C value\n",
    "  for i in cvals:\n",
    "    #our model\n",
    "    clf = LogisticRegression(solver='liblinear', C=i, random_state=1)\n",
    "    #accuracy score\n",
    "    score = cross_val_score(clf, X_train, y_train, fit_params={'sample_weight': train.instance_weights}, cv = 5)\n",
    "    #prediction labels for fairness\n",
    "    fair_pred = cross_val_predict(clf, X_train, y_train, fit_params={'sample_weight': train.instance_weights}, cv = 5)\n",
    "    test_pred = train.copy()\n",
    "    test_pred.labels = fair_pred\n",
    "\n",
    "    #defining our metric\n",
    "    metric = ClassificationMetric(train, test_pred, unprivileged_groups=unprivileged_groups, privileged_groups=privileged_groups)\n",
    "\n",
    "    #recording the scores\n",
    "    equ_opp.append(metric.equal_opportunity_difference())\n",
    "    f1score.append(f1_score(y_train, fair_pred, average = 'macro'))\n",
    "    avg_acc = np.mean(score)\n",
    "    avg_acc_scores.append(np.mean(score))\n",
    "\n",
    "  #taking the absolute value for the Equality of Opportunity scores\n",
    "  absol_equ_opp = []\n",
    "  for value in equ_opp:\n",
    "    absolute = abs(value)\n",
    "    absol_equ_opp.append(absolute)\n",
    "\n",
    "  #combining the values of accuracy and Equality of Opportunity\n",
    "  combined = [i + (1-j) for (i, j) in zip(avg_acc_scores , absol_equ_opp)]\n",
    "\n",
    "\n",
    "  #re-scaling the combined value to be out of 1, closer to 1 is better\n",
    "  rescaledcombined = []\n",
    "  for value in combined:\n",
    "    rescaled = value - 1\n",
    "    rescaledcombined.append(rescaled)\n",
    "  \n",
    "  best_c = equ_opp.index(min(equ_opp, key= abs))\n",
    "  best_acc_c = avg_acc_scores.index(max(avg_acc_scores))\n",
    "  best_combined = rescaledcombined.index(max(rescaledcombined)) # we want this to be \n",
    "  best_f1 = f1score.index(max(f1score)) #we want f1 score to be 1\n",
    "\n",
    "  print('Best accuracy score is at C: ', cvals[best_acc_c], 'with accuracy score of', max(avg_acc_scores))\n",
    "  print('Best Equality of Opportunity score is at C: ', cvals[best_c], 'with an equality of opportunity score of', min(equ_opp, key= abs))\n",
    "  print('Best f1 score is at C: ', cvals[best_f1], 'with a score of', max(f1score))\n",
    "  print('Best combined score is at C: ', cvals[best_combined], 'with a score of', max(rescaledcombined))\n",
    "  \n",
    "  #plotting the graph\n",
    "  xplt = cvals\n",
    "  plt.figure(figsize = (10,6))\n",
    "  plt.scatter(xplt, rescaledcombined, color = 'darkturquoise', label = 'Combined: Average and Equality of Opportunity')\n",
    "  plt.scatter(xplt, f1score, color = 'coral', label = 'F1 Score')\n",
    "  plt.legend(loc='upper center', bbox_to_anchor=(0.5, -0.10),\n",
    "          fancybox=True, shadow=True)\n",
    "\n",
    "  font1 = {'family':'serif','color':'darkturquoise','size':13}\n",
    "  font2 = {'family':'serif','color':'darkturquoise','size':11}\n",
    "  plt.title('Reweighed Scores', fontdict = font1, color = 'darkturquoise')\n",
    "  plt.xlabel('C Value', fontdict = font2)\n",
    "  plt.ylabel('Score', fontdict = font2)\n",
    "  plt.show()\n",
    "  \n",
    "\n",
    "  #return(df) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "Dims5wRq7WI6"
   },
   "source": [
    "### Adult"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 513
    },
    "id": "mY_elsViFIqT",
    "outputId": "b808a8ab-5674-4b7f-c93d-494bcd5e7f0b"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best accuracy score is at C:  0.05 with accuracy score of 0.7884409465674679\n",
      "Best Equality of Opportunity score is at C:  0.01 with an equality of opportunity score of -0.011180795108922947\n",
      "Best f1 score is at C:  0.005 with a score of 0.6794571444787936\n",
      "Best combined score is at C:  0.01 with a score of 0.776762930048776\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "rw_fairaccuracyf1(X_train_a, y_train_a, a_train, c)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 513
    },
    "id": "67QWkEzKFvny",
    "outputId": "8956b363-846b-485f-cb4b-d0a17eabd53e"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best accuracy score is at C:  0.014799999999999997 with accuracy score of 0.7884409465674679\n",
      "Best Equality of Opportunity score is at C:  0.004 with an equality of opportunity score of -0.0021185994499224314\n",
      "Best f1 score is at C:  0.004 with a score of 0.6797792820673395\n",
      "Best combined score is at C:  0.004 with a score of 0.7859421147027172\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "rw_fairaccuracyf1(X_train_a, y_train_a, a_train, np.arange(0.004, 0.06, 0.0009))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 513
    },
    "id": "P1J2paIqZx8d",
    "outputId": "81904079-9127-49a4-d1bd-d43c06d0e50f"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best accuracy score is at C:  0.001 with accuracy score of 0.7891137006660291\n",
      "Best Equality of Opportunity score is at C:  0.0031600000000000005 with an equality of opportunity score of -0.0021185994499224314\n",
      "Best f1 score is at C:  0.001 with a score of 0.6830599074062617\n",
      "Best combined score is at C:  0.0031600000000000005 with a score of 0.7859421147027172\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "rw_fairaccuracyf1(X_train_a, y_train_a, a_train, np.arange(0.001, 0.02, 0.00009))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "Lia8YLES7XkI"
   },
   "source": [
    "### German"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 513
    },
    "id": "zPFz_4lvEM7G",
    "outputId": "a885aee9-1695-4722-91ab-c3aaf3ce1c39"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best accuracy score is at C:  0.5 with accuracy score of 0.7128571428571429\n",
      "Best Equality of Opportunity score is at C:  0.01 with an equality of opportunity score of 0.0031311154598824276\n",
      "Best f1 score is at C:  0.005 with a score of 0.5257516199573378\n",
      "Best combined score is at C:  0.005 with a score of 0.7040900195694717\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "rw_fairaccuracyf1(X_train_g, y_train_g, g_train, c)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 513
    },
    "id": "j6hXBW9qZ858",
    "outputId": "810da4f8-c595-4232-ec85-5e2d5a669c3b"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best accuracy score is at C:  0.0019 with accuracy score of 0.7214285714285714\n",
      "Best Equality of Opportunity score is at C:  0.0271 with an equality of opportunity score of -0.000684931506849451\n",
      "Best f1 score is at C:  0.0019 with a score of 0.5726776674774834\n",
      "Best combined score is at C:  0.0046 with a score of 0.7063796477495106\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAmcAAAGrCAYAAACFcDBDAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3deZhU1YH///fpFsEtiAG3qID5qahQULJpFDFhXJIYlJgFQ4xLFJdhnCxjouNMNCaZaGLikq/jkqjESNToaIImM25RogZlsbDccA0q4oJgiKhgQ5/fH/d2WzTdUN1dRV2636/n6aerzl3qnK4u+sM5554bYoxIkiQpG+pqXQFJkiR9yHAmSZKUIYYzSZKkDDGcSZIkZYjhTJIkKUMMZ5IkSRmySa0rIGnjEgrFscAFwGhgHrAM6A30BM6N+dzvalCn7wA7x3zuX8rc/15gf+CwmM/dX4HXvws4cF3nC4Xi9sBPgf8PWAFsATwN/Czmc8XO1kFS12HPmaR2ifncDGBi+vRfYj53UMzn8sDNwPWhUBxYg2pdApxZ7s4xnxsHvF6pF4/53CFlnG8q8FbM5/aL+dwngbHAzsAnKlUPSV2D4UxSpdwC9ACGb+gXjvncypjPvbuhX7edDgDuaXoS87n3gXOBZ2tVIUnZ5LCmpEpp+vfk5aaCUCjuSzIECklwuzDmc7eGQnEC8N9AI/DfMZ/7USgUvwqcEvO5A0Kh2A/4E7AdcEzM52aEQvEE4GTgfWAV8K8xn3syPdc5wNYxnxuQvu5HgCuAwcAi4E7g58AMYGLM55p6uYaFQvF0YDeSIcZJMZ9rSM8xiKRHrlfatutjPnd5ui0A/0nSg/hqWtf1eQH4VigUCzGfWwQQ87m/lO4QCsU9gItJhjzrgWeA/4z53Ktpm34G7AkE4Cng2zGf+0coFM8A/hn4K7AEGEoSBncF3k7bviewmmQo+oyYz60IhWIv4Mq0/e8D/wDOivnc/DLaI6lK7DmT1GlpWPk8SZCYlZZ9jKSn6McxnxsLfBG4JhSKuZjP3Qb8EFgY87kfpacZD+wXCsVtYz63GLgc+Lc0mB0JXAhMiPncQSSh6c5QKPZKz/WNFlW6FOgD5GM+dxjJPC/SIdjS4cfhwFFAHsgBR6R13yKt+21p3T8D/FsoFD+XHncscApwYMznDiYJaDuu58c0mSQEvRwKxbtCoXhaKBS3LvkZbgncC9wS87kDgTFAP2BkustUYJOYzx0Q87n9Seb4XZu266fp9kNIAvCB6c93BfBrYHOSOXZjge2Bpp/5ccD2MZ/7RDrU+yCw73raIanKDGeSOuMXoVD8C/AWcDhwTcm2Y4DXYj73fwAxn3sV+DPw9XT7bcCIUCjuFArFzUh6g54lDUjAp/mwR+oU4H9KepxuJwkcn25ZoVAo1gNHA1NjPrc6Lb62jfr/T8znYsznVpH0KO2Wln8O2Ab4Vfp6y4A/ACem278K/CHmc2+l238HrGzrh5Tu8whJT9bnSXrzfgy8GArF/dNdDicJlFPT/RuBbwKzQ6G4HTCBpJeryeXA59NtTR6K+dxL6fHfS8uOAP5f2s5G4Dcl7XgbGBwKxc+GQnETktD723W1Q1L1OawpqTP+JeZzD6bDcbNJemtOSLftAvQNheL9Jft/lGTYjZjPLQqF4iyS0PEyydDjM8CEUCjeCMSYzy0vOdceLc61hCRAtdQX2BRYXFL2Vhv1/3vJ4xUkvVFNrxeAe0Kh+ULK3nw46f9jJEOIpZa08RrN0hA4HZgeCsUpwK0kvVgHpa+5pCRQEvO55wFCoTgqLXqz5HRvlNS16fHbLV5yl/T7haFQbAqPPYG3Q6HYO+ZzN4VCsSkEXgvcRDJc+8H62iKpegxnkjot5nPPhELxx8D3Q6H4vZjPLSQJXH9LhyEBCIViT5I5XE1uJelJehE4C9gJOIOk56t0HtfLwFMxn/tWybm2Ipl71tJbJOFi25Kyvu1s0svA+y3qvgnwkfTpqy3OD0nwbFMoFG+P+VzTsCgxn1seCsU/ACeVvOZHQ6FY3xTQQqG4I8ncs6Z5fNuR/KyaHlOyra12AJxculxHOnS8LBSKfYHfx3zu5lAo7kRyUcfP+TBgS6oBhzUlVcqlJD1RTQHqN8DH04sCmual/TfwhZJjbiWZW9U35nNvxnzuUZIhv/NIepiaXAEclYaJpvlZf+HDnqFmabD5LXB8OsQJyRBre9wOvBcKxS+WlJ1N0sPU1LYj0gsXCIXil0km8a/LkFAoTmp6kg7lTgDuKnnNt0nmgTUNz/4aGJDOk/sDyfBuk1NJgtUbtKHkuMklrzuWJIQBTCEdRk4D9bP4n3ap5kKMsdZ1kLQRaWUR2qtKrmL8FsnQ5iySOVSDSeZW1ZH0AN0NnBfzuVhyvseAa2M+d3H6/GfA0JjP/VOL1z2OJJysSM93Sczn/qfkas1BJHOuxoVCsTfJnKwhwEKSgHJ5zOdCeq4/AIcC80lCzsHAaem5L4j53OXp1ZoX8WHoeozk6sgP0qD5PeDLJEOdD5BcJPB34MymeXYt6n8C8BWSnsNVwFbpz+OcmM+tTPcpfc1NgGkxn7ss3bY1ydWag9JTzge+lfaAnQp8Nz33/BY9fr1JLqYYTHJF5jvAP8d8bmEoFEeTBOF6YEuSXsfJTXP7JNWG4UxSl5MGkvdKlsUYRTKBf4fa1kyS1s9hTUld0amkw4OprwN31KYqktQ+zi2Q1BU9BJyfzvHqRbLAbMu10CQpkxzWlCRJyhCHNSVJkjLEcCZJkpQhXWbOWd++feOAAQNqXQ1JkqT1mjt37lsxxn6tbesy4WzAgAHMmTOn1tWQJElarxDCS21tc1hTkiQpQwxnkiRJGWI4kyRJyhDDmSRJUoYYziRJkjLEcCZJkpQhhjNJkqQMMZxJkiRliOFMkiQpQwxnkiRJGWI4q6BpS99mwBNPU1coMuCJp5m29O1aV0mSJG1kusy9NWtt2tK3mfzyQt6LEYCXGhqY/PJCACZt06eWVZMkSRsRw1mFnL3o9eZg1uS9GDl70euGsxLTlr7N2Yte5+WGBrapq4MQWLp6Nbv06MGPdtzen5UkqdsznFXIyw0NrZa/1NDAtKVvd5nQ0Vq4WrJ6NfXAamj+/tFWtgWgNL4uaWxsfmxPoyRJCeecVcguPXq0uW3yywu7xPyzpqHblxoaiCThasnq1UASvkq/t7ZtzX7FtTX1NEqS1J3Zc1YhP9px+zXmnJVqbXizMz1Q5WwrZ5/+7RxKbG3ottLa6oGUJKm7MJxVSFPA+epLr7S6vTR0tLx4oHR4r7UeqI5sK2ef9g4lbojgtK4eSEmSuoOqDmuGEA4LITwTQng+hHBmK9svCiHMS7+eDSH8vWTbsSGE59KvY6tZz0qZtE0f+rcRLkpDx4bogSpXe4YSqx2cNg+BH+24fVVfQ5KkrKtaOAsh1AOXAZ8G9gKODiHsVbpPjPGbMcZhMcZhwC+AW9NjtwHOAUYDo4BzQggbxSzxH+24PZuHsEZZy9CRtaG7cuvTWts6Y4sQ+Gh9PYFkiPWqXXbyYgBJUrdXzWHNUcDzMcYXAUIINwJHAE+1sf/RJIEM4FDg7hjj0vTYu4HDgBuqWN+KaAoXTfPJWlsiYpcePXgpQwEtApsUiu2es9bRbe2d6yZ1JeXON63mZ7DWx3fXunXXdm+sdavl36oQqzS8FkL4AnBYjPHE9PkxwOgY45RW9u0PPAzsFGNcHUL4N6BXjPGH6fb/BN6PMV7Y1uuNGDEizpkzpxpNqbiWc842RpuHYE+X1E5d4bMvdSfV/FsXQpgbYxzR2rasLKUxEbglxrh6vXuWCCFMDiHMCSHMWbx4cZWqVnmTtunDVbvsRP8ePQgkif2j9fVAkuJLv3d0Wzn7dIbLXkjtl6X5ppLWr1Z/66o5rPkqsHPJ853SstZMBP65xbEHtTj2/pYHxRivAq6CpOes41Xd8CZt06emvU51heJ61x1bn6zNncuCpiGrlxoautXQgEMi5R0vaeNTi7911Qxns4HdQggDScLWROArLXcKIQwC+gAzS4rvBP6r5CKAQ4CzqljXbqcS895c9mJNLYesWi5d0vS9msujdMVzd6W6Sdr41OJvXdXCWYxxVQhhCknQqgeuiTE+GUI4D5gTY5ye7joRuDGWTH6LMS4NIfyAJOABnNd0cYAqY12L5pajkstetNbbtLH0hLTcJknqOmq1xFPVLgjY0DamCwKyoqNDcJW8gsUJ0uruPlpf3+WHc61b9273xlq3al+tua4LArxDQDdW63lv4ARpdW/9e/RgweA9a10NSRmTlas11U15UYG6K++IIakt9pxV2bqGDl2MNXsL8lZSdxoacEikfcf72Ze0LoazKlrf1XvtvfF4V9TZCxOyyAV6JUmd4bBmFZUzn6q7L+ZauiAvbJjFd6t5vPcIlSR1lj1n7VR6X7zW7ptZqtz5VN193lUWLkyQJCkr7Dlrh6ZhypcaGoh8OCw5benbre5f7sJ1LuYqSZKaGM7aobVhynUNS/5ox+3ZPIR1ntMrtiRJUinDWTu0NfzYVnlb86mcnyRJktrinLN2WNeyD32LT7J09eq15qE5n0qSJLWHPWft0NYwZQSWrF5d1jw0SZKkdTGctUPTMGX9evbr7stjSJKkjjOctdOkbfrQWMZ+3X15DEmS1DGGsw4oZ+kLl8eQJEkdYTjrgPUtkeHyGJIkqaMMZx1QukRG4MPb+QRcHkOSJHWOS2mUqbXbNi0YvGetqyVJkroYw1kZmm7b1HR3gKblMgB7yCRJUkU5rFmG9t62SZIkqaMMZ2Vo722bJEmSOspwVoa2lsVwuQxJklRphrMytLZ0hstlSJKkajCclaHl0hkulyFJkqrFqzXLNGmbPoYxSZJUdfacSZIkZYjhTJIkKUMMZ5IkSRliOJMkScoQw5kkSVKGGM4kSZIyxHAmSZKUIYYzSZKkDDGcSZIkZYjhTJIkKUMMZ5IkSRliOJMkScoQw5kkSVKGGM4kSZIyxHAmSZKUIYYzSZKkDDGcSZIkZYjhTJIkKUMMZ5IkSRliOJMkScoQw5kkSVKGGM4kSZIyxHAmSZKUIYYzSZKkDDGcSZIkZYjhTJIkKUMMZ5IkSRliOJMkScoQw5kkSVKGVDWchRAOCyE8E0J4PoRwZhv7fCmE8FQI4ckQwm9LyleHEOalX9OrWU9JkqSs2KRaJw4h1AOXAQcDC4HZIYTpMcanSvbZDTgL2D/G+HYIYduSU7wfYxxWrfpJkiRlUTV7zkYBz8cYX4wxfgDcCBzRYp+TgMtijG8DxBjfrGJ9JEmSMq+a4exjwCslzxemZaV2B3YPITwUQng4hHBYybZeIYQ5afmRVaynJElSZlRtWLMdr78bcBCwE/CXEMKQGOPfgf4xxldDCLsCfw4hPB5jfKH04BDCZGAywC677LJhay5JklQF1ew5exXYueT5TmlZqYXA9BhjQ4zxb8CzJGGNGOOr6fcXgfuBfMsXiDFeFWMcEWMc0a9fv8q3QJIkaQOrZjibDewWQhgYQtgUmAi0vOry9yS9ZoQQ+pIMc74YQugTQuhZUr4/8BSSJEldXNWGNWOMq0IIU4A7gXrgmhjjkyGE84A5Mcbp6bZDQghPAauBM2KMS0IInwCuDCE0kgTI80uv8pQkSeqqQoyx1nWoiBEjRsQ5c+bUuhqSJEnrFUKYG2Mc0do27xAgSZKUIYYzSZKkDDGcSZIkZYjhTJIkKUMMZ5IkSRliOJMkScoQw5kkSVKGGM4kSZIyxHAmSZKUIYYzSZKkDDGcSZIkZYjhTJIkKUMMZ5IkSRliOJMkScoQw5kkSVKGGM4kSZIyxHAmSZKUIYYzSZKkDDGcSZIkZYjhTJIkKUMMZ5IkSRliOJMkScoQw5kkSVKGGM4kSZIyxHAmSZKUIYYzSZKkDDGcSZIkZYjhTJIkKUMMZ5IkSRliOJMkScoQw5kkSVKGGM4kSZIyxHAmSZKUIYYzSZKkDDGcSZIkZYjhTJIkKUMMZ5IkSRliOJMkScoQw5kkSVKGGM4kSZIyxHAmSZKUIYYzSZKkDDGcSZIkZYjhTJIkKUMMZ5IkSRliOJMkScoQw5kkSVKGGM4kSZIyxHAmSZKUIYYzSZKkDDGcSZIkZYjhTJIkKUMMZ5IkSRlS1XAWQjgshPBMCOH5EMKZbezzpRDCUyGEJ0MIvy0pPzaE8Fz6dWw16ylJkpQVm1TrxCGEeuAy4GBgITA7hDA9xvhUyT67AWcB+8cY3w4hbJuWbwOcA4wAIjA3PfbtatVXkiQpC6rZczYKeD7G+GKM8QPgRuCIFvucBFzWFLpijG+m5YcCd8cYl6bb7gYOq2JdJUmSMqGa4exjwCslzxemZaV2B3YPITwUQng4hHBYO46VJEnqcqo2rNmO198NOAjYCfhLCGFIuQeHECYDkwF22WWXatRPkiRpg6pmz9mrwM4lz3dKy0otBKbHGBtijH8DniUJa+UcS4zxqhjjiBjjiH79+lW08pIkSbVQdjgLheJBoVD8WigUe4RCcb8yDpkN7BZCGBhC2BSYCExvsc/vSXrNCCH0JRnmfBG4EzgkhNAnhNAHOCQtkyRJ6tLKCmehUDwT+AFwNNAIfDkUit9a1zExxlXAFJJQ9TTwuxjjkyGE80II49Pd7gSWhBCeAu4DzogxLokxLk1fb3b6dV5aJkmS1KWVO+dsl5jPjQmF4uUxn1sNfCMUihev76AY45+AP7Uo+17J4wh8K/1qeew1wDVl1k+SJKlLKHdYc1n6PZaUbVbhukiSJHV75facbR4KxX8HdgmF4hdJ5oCtql61JEmSuqdye86+C/QCtgO+A7xOK0ORkiRJ6pxye87+C5gZ87nvrXdPSZIkdVi5PWefBe6pZkUkSZJUfjh7EHi/tCAUit+sfHUkSZK6t3KHNXsDT4VCcSawMi0bDVxUlVpJkiR1U+WGs0HA91uU7dzajpIkSeq4csPZSTGfm1lakPaiSZIkqYLKCmcxn5sZCsXDgM+kRX+K+dz/Va9akiRJ3VO599b8NvCfJPPNVgLfW9+9NSVJktR+Zc85i/nc/qUFoVD0vpeSJEkVVu5SGv9opeydSlZEkiRJ5fecbRYKxZ8BD6XP92/HsZIkSSpTuT1n3yZZhPbfgbOA94B/q1alJEmSuqtyr9Z8H/iP9ItQKNbHfG51NSsmSZLUHZV7teYvQ6F4R0nRb0OhOL5KdZIkSeq2yh3W7BHzucNLnk8EDGeSJEkVVm44e7P0ScznIrCs8tWRJEnq3sq94nK7UCj+C2terdm3OlWSJEnqvsrtOfsWMBq4C/g/YGRaJkmSpApaZzgLheLgUCh+Hng35nNfBb4EzCQZ0mzcAPWTJEnqVtbXc3YBMAIgFIp9gFuBx0nuGPCL6lZNkiSp+1nfnLMlMZ/7d4BQKB4DzIr5XNNaZ9dWu3KSJEndzfp6zkrvn3kYcEsb2yRJklQB6+s56xcKxZ7ATsAhwKkl2z5WtVpJkiR1U+sLZ78GXgK2As6P+dyboVAclpbPrHblMqk4A+6dBsvegt59YdwkyI2tda0kSVIXsc5wFvO5P4ZCcRegV8zn/pGWzQOGbojKZUpxBvzv1fB+yWjussVw++XJYwOaJEmqgPUuQhvzuQ+ADzZAXbKrOCMJYQ0r197WsDLpSTOcSZKkCih3Edru7d5prQezJsve2nB1kSRJXZrhrBzrC1+9vZOVJEmqDMNZOdYVvnr0TC4KkCRJqgDDWTnGTUpCWEubbQWfO9X5ZpIkqWLWe0GA+DB8uYSGJEmqMsNZuXJj1wxjxRlw0WTDmiRJqijDWUe0XFrD9c4kSVKFOOesI1pbWqNpvTNJkqROMJx1RFtLa7jemSRJ6iTDWUe0tbSG651JkqROMpx1RGtLa7jemSRJqgAvCOgIl9aQJElVYjjrqJZLa0iSJFWAw5qSJEkZYjiTJEnKEMOZJElShhjOJEmSMsRwJkmSlCGGM0mSpAwxnEmSJGWI4UySJClDDGeSJEkZYjiTJEnKEMOZJElShlQ1nIUQDgshPBNCeD6EcGYr248LISwOIcxLv04s2ba6pHx6NespSZKUFVW78XkIoR64DDgYWAjMDiFMjzE+1WLXm2KMU1o5xfsxxmHVqp8kSVIWVbPnbBTwfIzxxRjjB8CNwBFVfD1JkqSNXjXD2ceAV0qeL0zLWjoqhFAMIdwSQti5pLxXCGFOCOHhEMKRVaynJElSZtT6goDbgQExxhxwN/Drkm39Y4wjgK8AF4cQPt7y4BDC5DTAzVm8ePGGqbEkSVIVVW3OGfAqUNoTtlNa1izGuKTk6a+An5RsezX9/mII4X4gD7zQ4virgKsARowYEStY9/YpzoB7p8Gyt6B3Xxg3CXJja1YdSZK08apmz9lsYLcQwsAQwqbARGCNqy5DCDuUPB0PPJ2W9wkh9Ewf9wX2B1peSJANxRlw++WwbDEQk++3X56US5IktVPVes5ijKtCCFOAO4F64JoY45MhhPOAOTHG6cDpIYTxwCpgKXBceviewJUhhEaSAHl+K1d5ZsO906Bh5ZplDSuTcnvPJElSO1VzWJMY45+AP7Uo+17J47OAs1o57q/AkGrWrWKWvdW+ckmSpHWo9QUBG7/efdtXLkmStA6Gs84aNwl69FyzrEfPpFySJKmdqjqs2S00zSvzak1JklQBhrNKyI01jEmSpIpwWFOSJClDDGeSJEkZYjiTJEnKEMOZJElShhjOJEmSMsRwJkmSlCGGM0mSpAwxnEmSJGWI4UySJClDvENAJRVneBsnSZLUKYazSinOgNsvh4aVyfNli5PnYECTJEllc1izUu6d9mEwa9KwMimXJEkqk+GsUpa91b5ySZKkVhjOKqV33/aVS5IktcJwVinjJkGPnmuW9eiZlEuSJJXJCwIqpWnSv1drSpKkTjCcVVJurGFMkiR1isOakiRJGWLPWTW4GK0kSeogw1mluRitJEnqBIc1K83FaCVJUicYzirNxWglSVInGM4qzcVoJUlSJxjOKs3FaCVJUid4QUCluRitJEnqBMNZNbgYrSRJ6iCHNSVJkjLEcCZJkpQhhjNJkqQMMZxJkiRliOFMkiQpQwxnkiRJGWI4kyRJyhDDmSRJUoYYziRJkjLEcCZJkpQhhjNJkqQM8d6a1Vackd4EfTGEOoiNH37v3c+bokuSpDUYzqqpOANuvxwaVibPY+Oa35ctTraDAU2SJAEOa1bXvdM+DGZtaViZ7CdJkoThrLqWvVXZ/SRJUpdnOKum3n0ru58kSeryDGfVNG4S9Oi57n169Ez2kyRJwgsCqqtpkr9Xa0qSpDIZzqotN9bwJUmSyuawpiRJUoYYziRJkjLEcCZJkpQhhjNJkqQMqWo4CyEcFkJ4JoTwfAjhzFa2HxdCWBxCmJd+nViy7dgQwnPp17HVrKckSVJWVO1qzRBCPXAZcDCwEJgdQpgeY3yqxa43xRintDh2G+AcYAQQgbnpsW9Xq76SJElZUM2es1HA8zHGF2OMHwA3AkeUeeyhwN0xxqVpILsbOKxK9ZQkScqMaoazjwGvlDxfmJa1dFQIoRhCuCWEsHM7j5UkSepSan1BwO3AgBhjjqR37NftOTiEMDmEMCeEMGfx4sVVqaAkSdKGVM1w9iqwc8nzndKyZjHGJTHGlenTXwHDyz02Pf6qGOOIGOOIfv36VazikiRJtVLNcDYb2C2EMDCEsCkwEZheukMIYYeSp+OBp9PHdwKHhBD6hBD6AIekZZIkSV1a1a7WjDGuCiFMIQlV9cA1McYnQwjnAXNijNOB00MI44FVwFLguPTYpSGEH5AEPIDzYoxLq1VXSZKkrAgxxlrXoSJGjBgR58yZU+tqSJIkrVcIYW6McURr22p9QYAkSZJKGM4kSZIyxHAmSZKUIYYzSZKkDDGcSZIkZYjhTJIkKUOqts6Z1qM4A+6dBsvegs22TMreXw69+8K4SZAbW9v6SZKkmnCds1oozoDbL4eGlW3vEwLECJttlTx//x0IdRAbP/y+rm3l7NO7n0FQkqQaWNc6Z/ac1cK909YdzCAJZpCEquayxjW/r2tbOfssW5yERDCgSZKUEYazWlj2Vq1r8KGGlUlYNJxJG0arUxo62PtdzZ71ah7fXevWXdu9sdathqNLDmvWwkWTk16rzAhw7q21roTU9ZUzpUFSdvToCZ87tSoBzds3Zc24SckbnhW9+9a6BlL3UM6UBknZ0TS6tIE5rFkLTQm8dGhj1Qe1+Ue7R88kLEqqvixNaZBUnhp8bg1ntZIbu3Y3afNclMUbZhzeqzWlDat334xNaZC0XjUYXTKcZUlrgU1S1zFuknPOpI1JjUaXnHMmSRtKbmwyubh3PyAkPdtNvduhrvXv69qno9tqfXx3rVt3bffGWrfe/ap2McD62HMmSRuSPeSS1sNw1p21Nsct42u/SJLU1RnOuquW6y215w4D3llAkqSqcc5Zd9XZ9ZZqtPaLJEldnT1n3VUl1m1xzaa1dXSoeGO/BYq3fmnf8U4NkLQOhrPuqhLrLXlngTV1Zqi4nG2dPX5jPXdXrJtTAyStg8Oa3VVnbyHlnQXW5q151B5ODZDUBnvOuqs1biHl1ZoV4TCv2svfGUmtMJx1Z663VFnemkft5dQASa0wnKn2NvQ9Rat5vFQupwZIaoPhTLXV1iT6jW2Cd2vBLGvBcWM9d1esm1MDJK2D4UwbVnMv2Vuw2ZawYjnEWOtaVV7vfvDNq2pdC0nSRshwpg2nZS9ZaQ9UV+NEb0lSB7mUhjac7rTUhBO9JUkdZDjThtNdepOc6C1J6gTDmTac9vQmbbbVhxOpQ92a39e1rZx9qnl8737wuVOd6C1J6jDnnKl9Wk7oh/KvUFufHj0NNpKkbs9wpvKta0J/OctNtNSjJ2yyKby/POlVc2kBSZIMZ2qHSk/o3/wjLjchSVILzjlT+So9ob+7XCAgSVI7GM5UvkovD+FyE5IkrcVwpvKNm5TME6sEl5uQJKlVzjlT+Zom63f0ak3vKyhJ0noZztQ+ubGGKkmSqshhTUmSpAwxnEmSJGWI4UySJClDDGeSJEkZYjiTJEnKEMOZJElShhjOJEmSMsRwJuGHvxAAABtJSURBVEmSlCGGM0mSpAwxnEmSJGVIiDHWug4VEUJYDLxUpdP3Bd6q0rk3Brbf9nfX9nfntoPtt/3dt/0bou39Y4z9WtvQZcJZNYUQ5sQYR9S6HrVi+21/d21/d2472H7b333bX+u2O6wpSZKUIYYzSZKkDDGcleeqWlegxmx/99ad29+d2w623/Z3XzVtu3POJEmSMsSeM0mSpAzpluEshHBYCOGZEMLzIYQzW9neM4RwU7r9kRDCgJJtZ6Xlz4QQDi33nFnR0baHEA4OIcwNITyefv9UyTH3p+ecl35tu+Fa1D6daP+AEML7JW28ouSY4enP5fkQwqUhhLDhWtQ+nWj/pJK2zwshNIYQhqXbutL7f2AI4dEQwqoQwhdabDs2hPBc+nVsSflG8f53tO0hhGEhhJkhhCdDCMUQwpdLtk0NIfyt5L0ftqHa016dfO9Xl7Rxekn5wPRz8nz6udl0Q7SlIzrx/n+yxWd/RQjhyHRbV3r/vxVCeCr9Hb83hNC/ZNuG/+zHGLvVF1APvADsCmwKPAbs1WKf04Ar0scTgZvSx3ul+/cEBqbnqS/nnFn46mTb88CO6ePBwKslx9wPjKh1+6rc/gHAE22cdxawLxCA/wU+Xeu2Vrr9LfYZArzQRd//AUAOuA74Qkn5NsCL6fc+6eM+G8v738m27w7slj7eEXgN2Dp9PrV036x+dab96bblbZz3d8DE9PEVwKm1bms12l+yzzbAUmDzLvj+f7KkXafy4b/9Nfnsd8ees1HA8zHGF2OMHwA3Ake02OcI4Nfp41uAcWkiPgK4Mca4Msb4N+D59HzlnDMLOtz2GGMhxrgoLX8S2CyE0HOD1LpyOvPetyqEsAPwkRjjwzH5tF4HHFn5qldEpdp/dHrsxma97Y8xLogxFoHGFsceCtwdY1waY3wbuBs4bCN6/zvc9hjjszHG59LHi4A3gVYXzsywzrz3rUo/F58i+ZxA8rnJ4nsPlWv/F4D/jTG+V72qVkU57b+vpF0PAzulj2vy2e+O4exjwCslzxemZa3uE2NcBSwDPrqOY8s5ZxZ0pu2ljgIejTGuLCm7Nu3W/s+sDuvQ+fYPDCEUQggzQghjSvZfuJ5zZkWl3v8vAze0KOsq7397j91Y3v+K/BsVQhhF0vPwQknxj9KhoIsy/B+2zra/VwhhTgjh4aYhPZLPxd/Tz0lHzrkhVepv1ETW/ux3xff/6yQ9Yes6tqqf/e4YztQJIYS9gQuAk0uKJ8UYhwBj0q9jalG3KnsN2CXGmAe+Bfw2hPCRGtdpgwshjAbeizE+UVLcHd7/bi/tKfgNcHyMsal35SxgEDCSZNjnuzWqXrX1j8lq8V8BLg4hfLzWFdrQ0vd/CHBnSXGXe/9DCF8FRgA/rWU9umM4exXYueT5TmlZq/uEEDYBegNL1nFsOefMgs60nRDCTsBtwNdijM3/c44xvpp+fwf4LUkXchZ1uP3pUPYSgBjjXJKeg93T/XcqOT6r7z108v1PrfU/5y72/rf32I3l/e/Uv1Hpf0T+CJwdY3y4qTzG+FpMrASupWu+96W/4y+SzLHMk3wutk4/J+0+5wZWib9RXwJuizE2NBV0tfc/hPBPwNnA+JKRoZp89rtjOJsN7JZeZbMpyR+b6S32mQ40XZHxBeDP6ZjydGBiSK5oGwjsRjIhsJxzZkGH2x5C2JrkH+czY4wPNe0cQtgkhNA3fdwDOBx4gmzqTPv7hRDqAUIIu5K89y/GGF8D/hFC2Dcdzvsa8IcN0ZgO6MzvPiGEOpJ/oJvnm3XB978tdwKHhBD6hBD6AIcAd25E73+H257ufxtwXYzxlhbbdki/B5L5Nl3uvU/f857p477A/sBT6efiPpLPCSSfmyy+91CZv1FH0+I/Zl3p/Q8h5IErSYLZmyWbavPZr9SVBRvTF/AZ4FmS3o+z07Lz0jcFoBdwM8mE/1nAriXHnp0e9wwlV2a0ds4sfnW07cB/AO8C80q+tgW2AOYCRZILBS4B6mvdziq0/6i0ffOAR4HPlZxzBMk/Si8A/490cecsfnXyd/8g4OEW5+tq7/9Ikrkj75L0jDxZcuwJ6c/leZKhvY3q/e9o24GvAg0tPvvD0m1/Bh5P2389sGWt21mF9n8ibeNj6fevl5xz1/Rz8nz6uelZ63ZW6Xd/AEmvUF2Lc3al9/8e4I2S3/HpJcdu8M++dwiQJEnKkO44rClJkpRZhjNJkqQMMZxJkiRliOFMkiQpQwxnkiRJGbLJ+neRpI1HKBTrgO8AQ4HlJAvpNgCXxnzukZL9tia5cfPngHNjPveDtHx34BqS5QSOjfnc31ucfwvgKmD/mM8NqHZ7JHU/9pxJ6mouAz4S87mjYz53UsznvkSyvtHRpTuloWsiyf1DHy4pfxZ4EZjYMpil298lWe9QkqrCnjNJXUYoFAeQ3Ntz2xabLgb2abl/zOdWhELxtyQ3Or47Pcf2JIGtIRSKt5MsON0LWBLzuXNaec2vAT8GJpGsRH490Cfmcwel2z9Jsnr8aySLeZ6TBkBJapU9Z5K6kpHAyzGfe6+0MOZzy2I+d18bx1wNHBkKxY+mz48Ffp0+nhrzuX+L+dwUYEQoFEe3PDjmc9cBz6WP3yW5SwIA6TmvBf455nNnpa/1qw63TlK3YDiT1JW0+5YnMZ8rAE+R3KYIYHjM5+YAq4FdQqF4dSgULwYGktzsvj32AzYHfhYKxStIetfq21tHSd2Lw5qSupLZwM6hUNy8tPcsFIofAbaL+dxzbRx3NXByKBTnAg+mZRNJbmY8IuZzq0OhOJW2g1Xkw//s9mixbUnM504pqcuW7WmQpO7HnjNJXUbM514iuQLzzBabLgH+aR2HTgN2Ay4gmTMG8FFgWcznVqfPd1nH8a8DO6SPh5WUzwS2TefCNc1nu2mdjZDU7Xnjc0ldSrqUxmnAJ0gm9m8DzIz53MXrOW4asEnM576cPu8N3Jye4yVgHMnyGt8GvgVMAP4r5nPnh0JxP+DnwF+BD0h63M6N+dwvQ6E4DvhX4HmgD/CDmM+9WNlWS+pKDGeSJEkZ4rCmJElShhjOJEmSMsRwJkmSlCGGM0mSpAwxnEmSJGWI4UySJClDDGeSJEkZYjiTJEnKEO+tqW5t7ty5O9XV1d3V2Ng4CAi1ro+kjU6sq6ub39jYeMjw4cMX1roy6hoMZ+rW6urq7tp+++1322677UJdnR3JktqnsbExvPbaa3u8/PLLj4wfP3709OnTDWjqNP8aqVtrbGwctN12221iMJPUEXV1deywww51PXr02BE4a/z48dvVuk7a+PkXSd2dPWaSOqWuro4QAsDmwO41ro66AP8qSTX2+uuvM3HiRD7+8Y8zfPhwPvOZz/Dss892+rzHHXcct9xyy1rlc+bM4fTTT+/0+QGmTp3KlClTytr34osvplevXixbtqwir51lW265Zavl9fX1DBs2rPnr/PPPr/hrDxgwgLfeeguAT3ziEwAsWLCA3/72txV7jaOPPppcLsdFF1201rarrrqKQYMGMWjQIEaNGsWDDz5YsddtSyXad8UVV3DdddcBye/1okWLOnqqTTtVEQnnnEk1FWNkwoQJHHvssdx4440APPbYY7zxxhvsvnt1/gM+YsQIRowYUZVzr8sNN9zAyJEjufXWWzn++OM7fb7Vq1dTX19fgZptOJttthnz5s3bYK/317/+FfgwvHzlK1/p9Dlff/11Zs+ezfPPP7/WtjvuuIMrr7ySBx98kL59+/Loo49y5JFHMmvWLLbffvtOv3ZrVq1aVZH2nXLKKc2Pp06dyuDBg9lxxx0rUUWp3ew5k9ph2tK3GfDE09QVigx44mmmLX27U+e777776NGjxxp/GIYOHcqYMWOIMXLGGWcwePBghgwZwk033QTA/fffz9ixYzniiCPYddddOfPMM5k2bRqjRo1iyJAhvPDCC83nuueeexgxYgS77747d9xxR/Pxhx9+OADnnnsuJ5xwAgcddBC77rorl156afOx119/PaNGjWLYsGGcfPLJrF69GoBrr72W3XffnVGjRvHQQw+V1c4XXniB5cuX88Mf/pAbbrgBSHoqzjjjjOZ9Snvh2nrtLbfckm9/+9sMHTqUmTNnct555zFy5EgGDx7M5MmTiTECMHv2bHK5HMOGDWv+GUIS6M444wxGjhxJLpfjyiuvbLW+Rx55JMOHD2fvvffmqquuai7fcsstOfvssxk6dCj77rsvb7zxBgB/+9vf2G+//RgyZAj/8R//UdbPpNT//d//MWjQIPbZZx9OP/30Nd6fCy+8sHm/wYMHs2DBgnXWsVRTD96ZZ57JAw88wLBhw7jooos48MAD1wiJBxxwAI899tgax65YsYLjjz+eIUOGkM/nue+++wA45JBDePXVVxk2bBgPPPDAGsdccMEF/PSnP6Vv374A7LPPPhx77LFcdtllQNKr953vfIchQ4YwatSo5oB33HHHccopp6z1u9pWHaZOncr48eP51Kc+xbhx49ZqX8se3cMPP5z777+/+WfS2nvY9LO+5ZZbmDNnDpMmTWLYsGH88Y9/5Mgjj2w+1913382ECRPafjOlCjCcSWWatvRtJr+8kJcaGojASw0NTH55YacC2hNPPMHw4cNb3Xbrrbcyb948HnvsMe655x7OOOMMXnvtNSDpXbviiit4+umn+c1vfsOzzz7LrFmzOPHEE/nFL37RfI4FCxYwa9Ys/vjHP3LKKaewYsWKtV5n/vz53HnnncyaNYvvf//7NDQ08PTTT3PTTTfx0EMPMW/ePOrr65k2bRqvvfYa55xzDg899BAPPvggTz31VPN5pk+fzve+971W23LjjTcyceJExowZwzPPPMMbb7zBUUcdxW233da8z0033cTEiRPbfG2Ad999l9GjR/PYY49xwAEHMGXKFGbPns0TTzzB+++/3/xH/fjjj+fKK69sPr7J1VdfTe/evZk9ezazZ8/ml7/8JX/729/Wqu8111zD3LlzmTNnDpdeeilLlixpfv19992Xxx57jAMPPJBf/vKXAPzrv/4rp556Ko8//jg77LBDqz8DgPfff3+NYc2bbrqJFStWcNJJJ3H77bczd+5cXn/99TaPL6eOrTn//PMZM2YM8+bN45vf/CZf//rXmTp1KgDPPvssK1asYOjQoWscc9lllxFC4PHHH+eGG27g2GOPZcWKFUyfPp2Pf/zjzJs3jzFjxqxxzJNPPrnW7/OIESN48sknm5/37t2bxx9/nClTpvCNb3yjuby139W26gDw6KOPcssttzBjxoy12rcubb2HTb7whS8wYsQIpk2bxrx58/jMZz7D/PnzWbx4MZD85+SEE05Y52tInWU4k8p09qLXeS/tmWnyXoycvai8P6bt9eCDD3L00UdTX1/Pdtttx9ixY5k9ezYAI0eOZIcddqBnz558/OMf55BDDgFgyJAhzT0rAF/60peoq6tjt912Y9ddd2X+/Plrvc5nP/tZevbsSd++fdl222154403uPfee5k7dy4jR45k2LBh3Hvvvbz44os88sgjHHTQQfTr149NN92UL3/5y83nGT9+POedd16rbbnhhhuYOHEidXV1HHXUUdx8883069ePXXfdlYcffpglS5Ywf/589t9//zZfG5I5W0cddVTzee+77z5Gjx7NkCFD+POf/8yTTz7J3//+d9555x32228/gDWGuu666y6uu+46hg0bxujRo1myZAnPPffcWvW99NJLm3tWXnnlleZ9Nt100+ZereHDhzf/rB966CGOPvpoAI455pg239OmYc2mry9/+cvMnz+fgQMHsttuuxFC4Ktf/Wqbx5dTx3J88Ytf5I477qChoYFrrrmG4447bq19Hnzwwea6DBo0iP79+1dkLmTTz+noo49m5syZzeWt/a6uqw4HH3ww22yzTbtfv633sC0hBI455hiuv/56/v73vzNz5kw+/elPt/t1pfZwzplUppcbGtpVXo6999671Un769OzZ8/mx3V1dc3P6+rqWLVqVfO29AqyNp+3PFd9fT2rVq0ixsixxx7Lj3/84zX2/f3vf9/uuj7++OM899xzHHzwwQB88MEHDBw4kClTpjBx4kR+97vfMWjQICZMmEAIoc3XBujVq1dzT9iKFSs47bTTmDNnDjvvvDPnnntuqz2DpWKM/OIXv+DQQw9tc5/777+fe+65h5kzZ7L55ptz0EEHNZ+3R48ezT/Dpp9Vk9Z+tp21ySab0NjY2Py8qR7rqmM5Nt98cw4++GD+8Ic/8Lvf/Y65c+dWpL577bUXc+fO5VOf+lRz2dy5c9l7772bn5f+nNp63NrzlrbYYos2t7X1c4N1v4dtOf744/nc5z5Hr169+OIXv8gmm/inU9Vlz5lUpl169GhXeTk+9alPsXLlyjXmDBWLRR544AHGjBnDTTfdxOrVq1m8eDF/+ctfGDVqVLvOf/PNN9PY2MgLL7zAiy++yB577FHWcePGjeOWW27hzTffBGDp0qW89NJLjB49mhkzZrBkyRIaGhq4+eab13uuG264gXPPPZcFCxawYMECFi1axKJFi3jppZeYMGECf/jDH5p71tb12i01/cHt27cvy5cvbw65W2+9NVtttRWPPPIIQPOFFgCHHnool19+OQ1poH722Wd599131zjvsmXL6NOnD5tvvjnz58/n4YcfXm8b999//+bXaRqCLdegQYNYsGBB81zBpjl5kMzRevTRR4FkGK9pCLa9ddxqq61455131ig78cQTOf300xk5ciR9+vRZ65gxY8Y0t+XZZ5/l5ZdfXu/vz3e+8x2++93vNg+xzps3j6lTp3Laaac179M0d/Kmm25q7t2E1n9Xy61Dy/YNGDCAefPm0djYyCuvvMKsWbPWWe/1nW/HHXdkxx135Ic//GFFLmaR1sf4L5XpRztuz+SXF64xtLl5CPxox45fhRZC4LbbbuMb3/gGF1xwAb169WLAgAFcfPHFHHDAAcycOZOhQ4cSQuAnP/kJ22+/fatDk23ZZZddGDVqFP/4xz+44oor6NWrV1nH7bXXXvzwhz/kkEMOobGxkR49enDZZZex7777cu6557Lffvux9dZbM2zYsOZjpk+fzpw5c9Ya2rzxxhv505/+tEbZhAkTuPHGG/nud7/LnnvuyVNPPdUcPNt67f79+69xjq233pqTTjqJwYMHs/322zNy5MjmbVdffTUnnXQSdXV1jB07lt69ewNJIFmwYAH77LMPMUb69eu3Vm/gYYcdxhVXXMGee+7JHnvswb777rven9cll1zCV77yFS644AKOOOKINvdrmnNW+lrnn38+V111FZ/97GfZfPPNGTNmTHMwOOqoo7juuuvYe++9GT16dPMVvO2tYy6Xo76+nqFDh3LcccfxzW9+k+HDh/ORj3ykzbBx2mmnceqppzJkyBA22WQTpk6dukYva2vGjx/Pq6++yic+8QlCCGy11VZcf/31a8zDe/vtt8nlcvTs2XONINra72q5dWjZvm984xsMHDiQvfbaiz333JN99tlnnfVuqekChc0224yZM2ey2WabMWnSJBYvXsyee+7ZrnNJHRFiizk0Uncyd+7c2NaE/NZMW/o2Zy96nZcbGtilRw9+tOP2TNpm7V4H1dby5cubr1Q8//zzee2117jkkktqXKvy3H///Vx44YXNFzdUy6JFizjooIOYP38+G2oh5gEDBjBnzpzmqzmbHHfccRx++OF84Qtf2CD16IgpU6aQz+f5+te/3ur2uXPn8v3vf38qcP306dPv3aCVU5djz5nUDpO26WMY2wj88Y9/5Mc//jGrVq2if//+zVcmKnHddddx9tln8/Of/3yDBbON2fDhw9liiy342c9+VuuqqJuw50zdWnt7ziSpNfacqZL8L5MkSVKGGM4kSZIyxHAmSZKUIYYzSZKkDDGcSTVWX1+/xv0WFyxYwJIlS/jkJz/JlltuucYNnFu64447yOfzDB06lL322qvNG3lLkjYeLqUh1VjT/RZLvfvuu/zgBz/giSee4Iknnmj1uIaGBiZPnsysWbPYaaedWLly5XrvE7g+MUZijC6vIEk15L/AUnsUZ8BFk+HczyffizOq8jJbbLEFBxxwwDpX9H/nnXdYtWoVH/3oR4HkHplNt7Z54403mDBhAkOHDmXo0KH89a9/BeDnP/85gwcPZvDgwVx88cUALFiwgD322IOvfe1rDB48mFdeeYWf/vSnjBw5klwuxznnnFOVNkqSWmfPmVSu4gy4/XJoWJk8X7Y4eQ6QG9vh05be0mfgwIHcdtttZR23zTbbMH78ePr378+4ceM4/PDDOfroo6mrq+P0009n7Nix3HbbbaxevZrly5czd+5crr32Wh555BFijIwePZqxY8fSp08fnnvuOX7961+z7777ctddd/Hcc88xa9YsYoyMHz+ev/zlLxx44IEdbqMkqXz2nEnlunfah8GsScPKpLwTmoY1582bV3Ywa/KrX/2Ke++9l1GjRnHhhRdywgknAPDnP/+ZU089FUjmtPXu3ZsHH3yQCRMmsMUWW7Dlllvy+c9/ngceeACA/v37N9+f8a677uKuu+4in8+zzz77MH/+fJ577rlOtVGSVD57zqRyLXurfeUbyJAhQxgyZAjHHHMMAwcO7NCtirbYYovmxzFGzjrrLE4++eQK1lKSVC57zqRy9e7bvvIqW758Offff3/z83nz5tG/f38Axo0bx+WXJ0Ouq1evZtmyZYwZM4bf//73vPfee7z77rvcdtttjBkzZq3zHnrooVxzzTUsX74cgFdffZU333yz+g2SJAH2nEnlGzdpzTlnAD16JuVVMGDAAP7xj3/wwQcf8Pvf/5677rqLvfbaq3l7jJGf/OQnnHzyyWy22WZsscUWzb1ml1xyCZMnT+bqq6+mvr6eyy+/nP3224/jjjuOUaNGAXDiiSeSz+fXusLzkEMO4emnn2a//fYDYMstt+T6669n2223rUo7JUlr8sbn6tbafePz4oxkjtmyt5Ies3GTOnUxgKSuwRufq5LsOZPaIzfWMCZJqirnnEmSJGWI4UySJClDDGfq7mJjY2Ot6yBpI9bY2Ijzt1VJhjN1a3V1dfNff/311QY0SR3R2NjIa6+91rhixYq3gACY0tRpXhCgbq2xsfGQV1555aFFixbtEkKodXUkbWRijKxYsWLpdddd9xugH/CPWtdJGz/Dmbq14cOHLxw/fvwg4HRgD8AuNEkd0QeYCTxW64po4+c6ZxIwfvz4HsDHgJ61roukjdIKYOH06dNX17oi2vgZziRJkjLECwIkSZIyxHAmSZKUIYYzSZKkDPn/AQ5LDlhRq01CAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "rw_fairaccuracyf1(X_train_g, y_train_g, g_train, np.arange(0.001, 0.2, 0.0009) )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "o6jv482K-C0K"
   },
   "source": [
    "## Task 3a: Compute and Report the Final Scores"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "zqlTWMDz_OQK"
   },
   "source": [
    "### Adult"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "56ky20QY-pXp"
   },
   "source": [
    "<font color='darkviolet'> **(5a) Standard Logistic Regression Model**</font>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "8UN-mREE-EmU",
    "outputId": "7d9e0612-d511-4e00-ed19-84a1cf0c2240"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The final Accuracy score on the test data with hyperparameter C: 0.01 is 0.8039991810550741\n",
      "The final Equality of Opportunity score at C: 0.01 is [-0.4581379772270596]\n",
      "The disparate impact score is [0.0]\n"
     ]
    }
   ],
   "source": [
    "finalscores(X_train_a, y_train_a, X_test_a, y_test_a, a_test, 0.01)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "qjyBGV_CQdAs"
   },
   "source": [
    "> The <font color='darkviolet'>**final *accuracy* score**</font> in testing data at hyperparameter C: 0.01 is <font color='darkviolet'>**0.8040**</font> and the <font color='darkviolet'>**final *Equality of Opportunity* score**</font> is <font color='darkviolet'>**-0.45814**</font>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "4MdWHmid-7Li"
   },
   "source": [
    "\n",
    "\n",
    "<font color='darkturquoise'> **(6a) Fairness-based Logistic Regression Model**</font>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "WpFDbmv3-mFl",
    "outputId": "1de73776-d187-4fdd-82be-b4ed431307f4"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "After reweighing, the final accuracy score on the test data with hyperparameter C: 0.032 is 0.7909643076503106\n",
      "The Equality of Opportunity score at C: 0.032 is [0.02028996011236517]\n",
      "The disparate impact score is [0.6827554035209011]\n"
     ]
    }
   ],
   "source": [
    "rw_finalscores(X_train_a, y_train_a, X_test_a, y_test_a,a_train, a_test, 0.032)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "mtV1Bi4wSJx6"
   },
   "source": [
    "> The <font color='darkturquoise'>**final *accuracy* score**</font> in testing data at hyperparameter C: 0.032 is <font color='darkturquoise'>**0.79096**</font> and the <font color='darkturquoise'>**final *Equality of Opportunity* score**</font> is <font color='darkturquoise'>**-0.02029**</font>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "kX7RdeQm_gyi"
   },
   "source": [
    "### German"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "tlJn1c6L_g1g"
   },
   "source": [
    "<font color='darkviolet'> **(5a) Standard Logistic Regression Model**</font>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "o06frmKCAJpN",
    "outputId": "b5b79cf6-f080-4122-f277-1ebe9d2334e6"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The final Accuracy score on the test data with hyperparameter C: 0.051 is 0.6466666666666666\n",
      "The final Equality of Opportunity score at C: 0.051 is [-0.1504771225838023]\n",
      "The disparate impact score is [0.7754407051282051]\n"
     ]
    }
   ],
   "source": [
    "finalscores(X_train_g, y_train_g, X_test_g, y_test_g, g_test, 0.051)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "gqNSt06vPokX"
   },
   "source": [
    "> The <font color='darkviolet'>**final *accuracy* score**</font> in testing data at hyperparameter C: 0.051 is <font color='darkviolet'>**0.646667**</font> and the <font color='darkviolet'>**final *Equality of Opportunity* score**</font> is <font color='darkviolet'>**-0.150477**</font>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "-k2gYnlMAGUG"
   },
   "source": [
    "<font color='darkturquoise'> **(6a) Fairness-based Logistic Regression Model**</font>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "uupOvv7A-jW8",
    "outputId": "dcf553a8-ac1f-497c-d70d-69d0ddcc9401"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "After reweighing, the final accuracy score on the test data with hyperparameter C: 0.046 is 0.6366666666666667\n",
      "The Equality of Opportunity score at C: 0.046 is [-0.002936138977244984]\n",
      "The disparate impact score is [0.9815705128205129]\n"
     ]
    }
   ],
   "source": [
    "rw_finalscores(X_train_g, y_train_g, X_test_g, y_test_g,g_train, g_test, 0.046)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "Tg-b22zsRTLY"
   },
   "source": [
    "> The <font color='darkturquoise'>**final *accuracy* score**</font> in testing data at hyperparameter C: 0.046 is <font color='darkturquoise'>**0.636667**</font> and the <font color='darkturquoise'>**final *Equality of Opportunity* score**</font> is <font color='darkturquoise'>**-0.00294**</font>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "ObIR16SvBe6O"
   },
   "source": [
    "# **Task 3b (Extra)** "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "kpTzSsL2GLwO"
   },
   "source": [
    "This task will be using the metric *F1 Score* in the combined score to work out the optimal model. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "HVLNtHgCG9Ce"
   },
   "source": [
    "## (5b) Standard Machine Learning Model "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "mq0T_K_TIjJw"
   },
   "source": [
    "### Functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {
    "id": "ddKsHVAlCCnx"
   },
   "outputs": [],
   "source": [
    "#need to find a sweetspot between the fairness and accuracy\n",
    "#we want high accuracy and low equal opp \n",
    "#we would then want high 1-(low equal opp)\n",
    "#add (accuracy and high 1-(low equal opp)) - 1 would be the best combined score\n",
    "\n",
    "#looking into the f1 score \n",
    "\n",
    "def fairf1combined(X_train, y_train, train, cvals):\n",
    "  #empty lists for our scores\n",
    "  fairness = []\n",
    "  equ_opp = []\n",
    "  f1score = []\n",
    "\n",
    "#looping through each C value\n",
    "  for i in cvals:\n",
    "    #our model\n",
    "    clf = LogisticRegression(solver='liblinear', C=i, random_state=1)\n",
    "    #accuracy score\n",
    "    score = cross_val_score(clf, X_train, y_train, cv = 5)\n",
    "    #prediction labels for fairness\n",
    "    fair_pred = cross_val_predict(clf, X_train, y_train, cv = 5)\n",
    "    test_pred = train.copy()\n",
    "    test_pred.labels = fair_pred\n",
    "\n",
    "    #defining our metric\n",
    "    metric = ClassificationMetric(train, test_pred, unprivileged_groups=unprivileged_groups, privileged_groups=privileged_groups)\n",
    "\n",
    "    #recording the scores\n",
    "    equ_opp.append(metric.equal_opportunity_difference())\n",
    "    f1score.append(f1_score(y_train, fair_pred, average = 'macro'))\n",
    "\n",
    "  #taking the absolute value for the Equality of Opportunity scores\n",
    "  absol_equ_opp = []\n",
    "  for value in equ_opp:\n",
    "    absolute = abs(value)\n",
    "    absol_equ_opp.append(absolute)\n",
    "\n",
    "  #combining the values of f1 scoreand Equality of Opportunity\n",
    "  combined = [(1-j) + k for (j, k) in zip(absol_equ_opp, f1score)]\n",
    "\n",
    "\n",
    "  #re-scaling the combined value to be out of 1, closer to 1 is better\n",
    "  rescaledcombined = []\n",
    "  for value in combined:\n",
    "    rescaled = value - 1\n",
    "    rescaledcombined.append(rescaled)\n",
    "  \n",
    "  best_c = equ_opp.index(min(equ_opp, key= abs)) #we want this to be closest to 0\n",
    "  best_combined = rescaledcombined.index(max(rescaledcombined)) # we want this to be 1\n",
    "  best_f1 = f1score.index(max(f1score)) #we want f1 score to be 1\n",
    "\n",
    "  #printing the values\n",
    "  print('Best Equality of Opportunity score is at C: ', cvals[best_c], 'with an equality of opportunity score of', min(equ_opp, key= abs))\n",
    "  print('Best f1 score is at C: ', cvals[best_f1], 'with a score of', max(f1score))\n",
    "  print('Best combined score is at C: ', cvals[best_combined], 'with a score of', max(rescaledcombined))\n",
    "  \n",
    "  #plotting the graph\n",
    "  xplt = cvals\n",
    "  plt.figure(figsize = (10,6))\n",
    "  plt.scatter(xplt, rescaledcombined, color = 'darkgreen', label = 'Combined: Average, F1 and Equality of Opportunity Scores')\n",
    "  plt.legend(loc='upper center', bbox_to_anchor=(0.5, -0.10),\n",
    "          fancybox=True, shadow=True)\n",
    "\n",
    "  font1 = {'family':'serif','color':'darkgreen','size':13}\n",
    "  font2 = {'family':'serif','color':'darkgreen','size':11}\n",
    "  plt.title('Combined Scores', fontdict = font1, color = 'darkviolet')\n",
    "  plt.xlabel('C Value', fontdict = font2)\n",
    "  plt.ylabel('Score', fontdict = font2)\n",
    "  plt.show()\n",
    "  \n",
    "\n",
    "  #return(df) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "0YapvbZAIGwH"
   },
   "source": [
    "### Adult "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 481
    },
    "id": "6Pe4jkfSFLmz",
    "outputId": "84f13512-3c00-4bba-fccd-77f329c02c2d"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best Equality of Opportunity score is at C:  0.001 with an equality of opportunity score of -0.4464800923254472\n",
      "Best f1 score is at C:  0.017 with a score of 0.6829809555455333\n",
      "Best combined score is at C:  0.001 with a score of 0.23314021908236526\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fairf1combined(X_train_a, y_train_a, a_train, np.arange(0.001, 0.03, 0.002))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "i9RdMh6DIWHl"
   },
   "source": [
    "### German"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 481
    },
    "id": "7KeJHLf3IYNm",
    "outputId": "bfe01374-723f-4d41-96cb-b845d5eaf7e3"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best Equality of Opportunity score is at C:  0.05 with an equality of opportunity score of -0.050097847358121395\n",
      "Best f1 score is at C:  0.005 with a score of 0.5569887959190813\n",
      "Best combined score is at C:  0.005 with a score of 0.49250738691712437\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fairf1combined(X_train_g, y_train_g, g_train, c)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 481
    },
    "id": "WvXryyudIyvN",
    "outputId": "5f14f38e-ce8c-4674-9516-348fa4768fd8"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best Equality of Opportunity score is at C:  0.051000000000000004 with an equality of opportunity score of -0.050097847358121395\n",
      "Best f1 score is at C:  0.001 with a score of 0.5719406041986688\n",
      "Best combined score is at C:  0.003 with a score of 0.5159089955736154\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAmcAAAGcCAYAAACcDwdWAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOzdeXxddZ3/8dcnhQIBBDoge5rCVKCki5LWFkQosilY7OACFKQ4tC5hcGXA6QgVqIIwDjJTl2aGZTAsws9KXXFQEFEKTR3EgrJVWpYqhQICtXT7/P64N/E2pG1Kc5PT9PV8PO6j937P9vnek5u8+z3n3BOZiSRJkoqhprcLkCRJ0t8YziRJkgrEcCZJklQghjNJkqQCMZxJkiQViOFMkiSpQLbo7QIkbfqmx7y9gS8CBwDLga2Ah4BvN2XDz6uwvYnAecCfmrLh8LXMcwjwH03Z8Lbu3n55/T8F3gkc25QNd65lnt2Ay4C/B5YB2wK/B/6tKRseqEZdkjZ9jpxJ2ijTY149cB9wd1M2jGnKhsOAw4FVwHersc2mbLgGuGQ9s/0aeFc1tl+u4WjgT+uZ7RrgufL7MhY4DNgbOLhadUna9BnOJG2s/wRmN2XDVW0NTdmwDPgn4C+9VVRTNmRTNrzQW9svewdwe9uLpmz4KzAVeKS3CpJUfB7WlPSGTY95OwHvBiZ3nNaUDcumx7yGinl3B64EdqX0u+eXwBeasmH59Jj3b8AE4CZgR2AI8CJwKqXDlyOB/sAHmrJhQYcavgSMAXYBvtaUDc3TY94g4GpKI1WDmrLhiekx72fAIeX1vRMYTOkQ44SmbFhRXtf+wNeArcs1frspG75RnhbAF4CTgKeBH3XhLXoc+Mz0mPd/TdnwTPl9uatD/fsBV1A65NkPeLj8vjw9Pea9Cfg3SoeLg9Kh4s82ZcNfpse8c4AmSiOEzwPDKYXBfYAXgK+Wl1sF3A+cU94nWwPfKvf/r5QC9OebsuEPXeiPpB7gyJmkjfEWSr9HnuxsYlM2/AVgesyrAWYB85qy4Z2UQlMj8OXyfJ8FfkLpMGRTUzaMpHTu2o+AS5qy4R3Ab4DPddjEaGBW+ZDheOA/pse8dzRlwx87novWlA3vonQY8iDgROCtwDDghHKN21Ia5ZpZPjT7HuBz02Pee8urOB34GPDOpmw4ilJA22M9789kSiFo4fSY99PpMe8T02Pejm0Tp8e87YCfAbeU35dDKYXMkeVZrgG2aMqGdzRlwyGUzuW7utyfy8rTjwYuLy9/MaVz264FaimF0cOA3YBp5XVOBHZryoaDy+/J3eX3UVJBGM4kbYyu3px3JKUw9i2A8kjVfwEfK49Itbm9KRteKT//P+DJpmz4c/n1XEphsNKjTdkwu7zOR4E7KYWodflu+ZDnSkojSoPL7e8FBpTroikbXgJuBc4sTz8VuLUpG54rT/8O8Nq6NtSUDfdSGsn6B+AZSmF0fvliBYDjgZ0ohSyasmE18GlgzvSYtyulwPmtilV+A/iH8rQ2v2obTWzKhvPLbScA/1nu52rguop+vAA0TI95x02PeVtQGim8fl39kNSzPKwpaWM8CqymdJL7ugws//tsRdufKY3u7AwsLre9WDF9ZYfXKyiNHFV6vsPrZ4E911NL5XloyyrWWUfp0OHt02Ne2/Qd+NtJ/3tSOoS4ru2/TjkEzgJmTY95Z1G6SGIapYsm6oDnm7JhVcX8jwFMj3mjKvrUpi2o1lU873heXV3538unx7y28LgV8ML0mLdDUzbcND3mtYXAqykdSv4CpZFKSQXgyJmkN6x8wv0PKI0ArWF6zNtzesy7vXy4cGG5uXLEZ1dgKfDcRpTwdx1e7wo89QbXtRD4a1M2HN72oDTiN6E8/WngzevZ/hqmx7zvV74ujwreSum8urZt/t30mNevYpk9yl9Nsrb3jIppa+sHwEcr+jIGGNWUDS9Nj3k7A99ryoYjgbeV+/jVdfVDUs8ynEnaWGcBI6fHvDPaGqbHvO0pHY6b3ZQNr1L6qo3fAB8vT9+S0mG2GU3Z0NVDo50ZMj3mjS6vcz9Ko1H/8wbX9X1g6fSY94GKtimURpigdGjwhOkxb5fy9j5E6ST+dRk6Pea1hTumx7xtKB2q/GnFNl+gdB4Y5ZB2LVDflA1/ohTkPlaxvo9TClZ/Zi0qlmu/SGN6zDsMuKX88izK59k1ZcNTlK4c9SiKVCCRuTG/FyWpNEoGfAnYn9JoWH+gpSkbvl4xzx787WrNLSldrfmvTdnw2vSY90Xgo5QOM15aXuRcSldNfovSuWFfpnRi+y2UTmI/j9Ko2/3AUEon0v97J1dr3gtMonSy/DHAHyiFnKOAT7RtsykbvlG+WvPf+Vvo+i2lqyOXl8+NOx/4EKVDnb+kdH7bi8B5Tdnwk07el48Ap5T7sRLYHvhf4IKmbHitPE/lNrcov2/Ty9N2pHS15v7lVf4B+Ex5BOzjFe/RHyovgJge83YALgcaKF2R+TKlCy2emh7z3g5cSOnK0O3K7+HktqtJJfU+w5kkSVKBeFhTkiSpQAxnkiRJBWI4kyRJKhDDmSRJUoEYziRJkgqkz3y3zc4775z19fW9XYYkSdJ6zZ0797nM3KWzaX0mnNXX19Pa2trbZUiSJK1XRCxY27SqHtaMiGMj4uGIeCwizutk+sSIWBwR95cfZ5bbR0TEPRHxYEQ8EBEfqmadkiRJRVG1kbOI6AdMp/Qt3E8BcyJiVmY+1GHWmzLzrA5tS4EPZ+ajEbEHMDcibsvMF5EkSerDqjlyNgp4LDPnZ+Zy4EbK93Nbn8x8JDMfLT9/BniW0q1ZJEmS+rRqhrM9gScrXj9VbuvoxPKhy1siYu+OEyNiFKX79D1enTIlSZKKo7e/SuP7QH1mDqN0M+BrKydGxO7AdcAZmbm648IRMTkiWiOidfHixT1SsCRJUjVVM5w9DVSOhO1VbmuXmc9n5mvll/8FHNQ2LSLeBPwQmJKZszvbQGbOyMzGzGzcZRePekqSpE1fNcPZHGBwRAyKiP7AScCsyhnKI2NtxgG/L7f3B2YC/5OZt1SxRkmSpEKp2tWambkyIs4CbgP6AVdl5oMRcSHQmpmzgLMjYhywElgCTCwv/kHgncDfRURb28TMvL9a9UqSJBVBZGZv19AtGhsb0y+hlSRJm4KImJuZjZ1N6+0LAiRJklTBcLYRWma3UH9uPTWTaqg/t56W2S29XZIkSdrE9Zl7a/a0ltktTL5uMkuXLwVgwZIFTL5uMgATRk/ozdIkSdImzJGzN2jKzCntwazN0uVLmTJzSi9VJEmS+gLD2Ru0cMnCDWqXJEnqCsPZG1Q3oG6D2iVJkrrCcPYGTRs/jdr+tWu01favZdr4ab1UkSRJ6gsMZ2/QhNETmHHaDAYOGEgQDBwwkBmnzfBiAEmStFH8ElpJkqQe5pfQSpIkbSIMZ5IkSQViOJMkSSoQw5kkSVKBGM4kSZIKxHAmSZJUIIYzSZKkAjGcSZIkFYjhTJIkqUAMZ5IkSQViOJMkSSoQw5kkSVKBGM4kSZIKxHAmSZJUIIYzSZKkAjGcSZIkFYjhTJIkqUAMZ5IkSQViOJMkSSoQw5kkSVKBGM4kSZIKxHAmSZJUIIYzSZKkAjGcSZIkFYjhTJIkqUAMZ5IkSQVS1XAWEcdGxMMR8VhEnNfJ9IkRsTgi7i8/zqyYdnpEPFp+nF7NOiVJkopii2qtOCL6AdOBo4CngDkRMSszH+ow602ZeVaHZQcAFwCNQAJzy8u+UK16JUmSiqCaI2ejgMcyc35mLgduBE7o4rLHAP+bmUvKgex/gWOrVKckSVJhVDOc7Qk8WfH6qXJbRydGxAMRcUtE7L2By0qSJPUpvX1BwPeB+swcRml07NoNWTgiJkdEa0S0Ll68uCoFSpIk9aRqhrOngb0rXu9VbmuXmc9n5mvll/8FHNTVZcvLz8jMxsxs3GWXXbqtcEmSpN5SzXA2BxgcEYMioj9wEjCrcoaI2L3i5Tjg9+XntwFHR8ROEbETcHS5TZIkqU+r2tWambkyIs6iFKr6AVdl5oMRcSHQmpmzgLMjYhywElgCTCwvuyQiLqIU8AAuzMwl1apVkiSpKCIze7uGbtHY2Jitra29XYYkSdJ6RcTczGzsbFpvXxAgSZKkCoYzSZKkAjGcSZIkFYjhTJIkqUAMZ5IkSQViOJMkSSoQw5kkSVKBGM4kSZIKxHDWzVpmt1B/bj01k2qoP7eeltktvV2SJEnahFTt9k2bo5bZLUy+bjJLly8FYMGSBUy+bjIAE0ZP6M3SJEnSJsKRs240ZeaU9mDWZunypUyZOaWXKpIkSZsaw1k3Wrhk4Qa1S5IkdWQ460Z1A+o2qF2SJKkjw1k3mjZ+GrX9a9doq+1fy7Tx03qpIkmStKkxnHWjCaMnMOO0GQwcMJAgGDhgIDNOm+HFAJIkqcsiM3u7hm7R2NiYra2tvV2GJEnSekXE3Mxs7GyaI2eSJEkFYjiTJEkqEMOZJElSgRjOJEmSCsRwJkmSVCCGM0mSpAIxnEmSJBWI4UySJKlADGeSJEkFYjiTJEkqEMOZJElSgRjOJEmSCsRwJkmSVCCGM0mSpAIxnEmSJBWI4UySJKlADGeSJEkFYjiTJEkqkKqGs4g4NiIejojHIuK8dcx3YkRkRDSWX28ZEddGxO8i4vcR8flq1ilJklQUVQtnEdEPmA68GxgCnBwRQzqZb3vgk8C9Fc0fALbKzKHAQcBHI6K+WrVKkiQVRTVHzkYBj2Xm/MxcDtwInNDJfBcBlwLLKtoS2DYitgC2AZYDf6lirZIkSYVQzXC2J/Bkxeunym3tIuJtwN6Z+cMOy94CvAosAhYCl2fmkirWKkmSVAi9dkFARNQAXwU+28nkUcAqYA9gEPDZiNink3VMjojWiGhdvHhxVeuVJEnqCdUMZ08De1e83qvc1mZ7oAG4MyKeAEYDs8oXBZwC/CQzV2Tms8CvgMaOG8jMGZnZmJmNu+yyS5W6IUmS1HOqGc7mAIMjYlBE9AdOAma1TczMlzJz58ysz8x6YDYwLjNbKR3KPAIgIralFNz+UMVaJUmSCqFq4SwzVwJnAbcBvwe+k5kPRsSFETFuPYtPB7aLiAcphbyrM/OBatUqSZJUFJGZvV1Dt2hsbMzW1tbeLkOSJGm9ImJuZr7ulC3wDgGSJEmFYjiTJEkqEMNZD2iZ3UL9ufXUTKqh/tx6Wma39HZJkiSpoLbo7QL6upbZLUy+bjJLly8FYMGSBUy+bjIAE0ZP6M3SJElSATlyVmVTZk5pD2Ztli5fypSZU3qpIkmSVGSGsypbuGThBrVLkqTNm+GsyuoG1G1QuyRJ2rwZzqps2vhp1PavXaOttn8t08ZP66WKJElSkRnOqmzC6AnMOG0GAwcMJAgGDhjIjNNmeDGAJEnqlHcIkCRJ6mHeIUCSJGkTYTiTJEkqEMOZJElSgRjOJEmSCsRwJkmSVCCGM0mSpAIxnEmSJBWI4UySJKlADGeSJEkFYjiTJEkqEMOZJElSgRjOeknL7Bbqz62nZlIN9efW0zK7pdM2SZK0edmitwvYHLXMbmHydZNZunwpAAuWLOCMa84gCJavWt7eNvm6yQBMGD2h12qVJEk9y5GzXjBl5pT2YNZmxaoV7cGszdLlS5kyc0pPliZJknqZ4awXLFyysCrzSpKkTZ/hrBfUDairyrySJGnTZzjrBdPGT6O2f+0abVv225L+/fqv0Vbbv5Zp46f1ZGmSJKmXGc56wYTRE5hx2gwGDhhIEAwcMJCrJ17NVROvWqNtxmkzvBhAkqTNTGRmb9fQLRobG7O1tbW3y5AkSVqviJibmY2dTXPkTJIkqUAMZ5IkSQViOJMkSSoQw5kkSVKBGM4kSZIKxHAmSZJUIFUNZxFxbEQ8HBGPRcR565jvxIjIiGisaBsWEfdExIMR8buI2LqatUqSJBXBFtVacUT0A6YDRwFPAXMiYlZmPtRhvu2BTwL3VrRtAXwbOC0zfxsRfwesqFatkiRJRVHNkbNRwGOZOT8zlwM3Aid0Mt9FwKXAsoq2o4EHMvO3AJn5fGauqmKtkiRJhVDNcLYn8GTF66fKbe0i4m3A3pn5ww7LvgXIiLgtIn4TEf9cxTolSZIKo2qHNdcnImqArwITO5m8BfAOYCSwFPhZ+TYHP+uwjsnAZIC6urqq1itJktQTqjly9jSwd8XrvcptbbYHGoA7I+IJYDQwq3xRwFPAXZn5XGYuBX4EvK3jBjJzRmY2ZmbjLrvsUqVuSJIk9ZxqhrM5wOCIGBQR/YGTgFltEzPzpczcOTPrM7MemA2My8xW4DZgaETUli8OOAx46PWbkCRJ6luqFs4ycyVwFqWg9XvgO5n5YERcGBHj1rPsC5QOec4B7gd+08l5aZIkSX1OZGZv19AtGhsbs7W1tbfLkCRJWq/yufSNnU3zDgGSJEkFYjiTJEkqEMOZJElSgRjOJEmSCsRwJkmSVCCGM0mSpAIxnEmSJBVIl++tGZPicKAOuAFozOa8p1pFSZIkba66NHIWk+I84CLgZGA18KGYFJ+pZmGSJEmbo64e1qzL5jwUeCKbc1U256cojaJJkiSpG3U1nL1U/rfyXk/bdHMtkiRJm72unnNWG5PiX4C6mBQfAI4GVlavLKlYWma3MGXmFBYuWUjdgDqmjZ8G0K1tE0ZP6JHtWGPvt20KNW6qdVujNVZjOz2tSzc+j0mxNfAvwLvLTT8BvpTN+dcq1rZBvPG5qqVldguTr5vM0uVL29u27LclQbB81fJuaavtX8vpY07n2nuurep2rNEa+3Ld1miN1djOjNNmVCWgrevG510NZ18F7snmvLm7i+suhjNVS/259SxYsqDq2+lX049Vq1dVfTsbwxq7x6ZQY2c2hbqtsXtY498MHDCQJy59otvXu65w1tVzzo4Dbu++kqRNx8IlC3tkO0X/RQjW2F02hRo7synUbY3dwxr/pqf+BlTqaji7G1jjEGZMik93fzlS8dQN6JkLk/vV9OuR7WwMa+wem0KNndkU6rbG7mGNf9NTfwMqdTWc7QA8FJOiJSbFVTEprgLOrGJdUmFMGz+N2v61a7Rt2W9L+vfr321ttf1rmXzo5Kpvxxp7v21TqHFTrdsarbEa22m7UKAndTWc7Q98Efgp8Ivy45lqFSUVyYTRE5hx2gwGDhhIEAwcMJCrJ17NVROv6ra2GafN4Ounfr3q27HG3m/bFGrcVOu2RmusxnaKfLXmmI63a4pJ8ZZszkeqVtkG8oIASZK0qVjXBQFd+p6zbM57YlIcC7yn3PSjbM6fdFeBkiRJKunqvTU/C3wBeK38ON97a0qSJHW/rt4hYP9szkMqG8oXBUiSJKkbdfWCgL900vZydxYiSZKkro+cbROT4t+AX5VfH7IBy0qSJKmLujpy9llKX0L7L8DngaXA56pVlCRJ0uaqq1dr/hX41/KDmBT9sjmLf28HSZKkTUxXr9Zsjknxg4qm62NSjKtSTZIkSZutrh7W3DKb8/iK1ycBhjNJkqRu1tVw9mzli2zOBF7q/nIkSZI2b1294nLXmBT/xJpXa+5cnZIkSZI2X10dOfsM8HZKNz7/CTCy3CZJkqRutM5wFpOiISbFPwCvZnOeCnwQuIfSIc3VPVCfJEnSZmV9I2eXAo0AMSl2Ar4L/I7SHQP+o7qlSZIkbX7Wd87Z89mc/wIQk+I04L5szrbvOru62sVJkiRtbtY3clZ5/8xjgVvWMk2SJEndYH3hbJeYFFvFpNgXOBr4XsW0Pde38og4NiIejojHIuK8dcx3YkRkRDR2aK+LiFciwltFSZKkzcL6DmteCywAtgcuyeZ8NibFiHL7PetaMCL6AdOBo4CngDkRMSszH+ow3/bAJ4F7O1nNV4Efd6UjkiRJfcE6w1k25w9jUtQBW2dz/qXcdj8wvAvrHgU8lpnzASLiRuAE4KEO811E6cKDcyobI+J9wB+BV7uwLQEts1uYMnMKC5cspG5AHdPGTwOoetuE0RN6bds9VeOE0ROqtdskSVpDZGZ1VhzxfuDYzDyz/Po04O2ZeVbFPG8DpmTmiRFxJ/C5zGyNiO2A/6U06vY54JXMvLyTbUwGJgPU1dUdtGDBgqr0ZVPQMruFyddNZunype1tW/bbkiBYvmp51dpq+9dy+pjTufaea3t82z1VY23/WmacNsOAJknqNhExNzMbO53WW+EsImqAnwMTM/OJDuHscuC+zPxORExlLeGsUmNjY7a2tlalL5uC+nPrWbCkd8Jpv5p+rFq9qle23VUbW+PAAQN54tInuq8gSdJmbV3hrKu3b3ojngb2rni9V7mtzfZAA3BnRADsBsyKiHGU7kbw/oj4CrAjsDoilmXmf1ax3k3awiULe23bRQ9msPE19ub7K0navFQznM0BBkfEIEqh7CTglLaJmfkSFffnrBw5Aw6taJ9KaeTMYLYOdQPqHDlbh42tsW5AXTdWI0nS2nX13pobLDNXAmcBtwG/B76TmQ9GxIXl0TF1o2njp1Hbv3aNti37bUn/fv2r2lbbv5bJh07ulW33VI21/WvbLxSQJKnaqhbOADLzR5n5lszcNzOnldvOz8xZncx7eHnUrGP71PWdbyaYMHoCM06bwcABAwmCgQMGcvXEq7lq4lVVbZtx2gy+furXe2XbPVWjFwNIknpS1S4I6Gmb+wUBkiRp07GuCwKqOnImSZKkDWM4kyRJKhDDmSRJUoEYziRJkgrEcCZJklQghjNJkqQCMZxJkiQViOFMkiSpQAxnkiRJBWI4kyRJKhDD2SagZXYL9efWUzOphvpz62mZ3dJpmyRJ2vRt0dsFaN1aZrcw+brJLF2+FIAFSxZwxjVnEATLVy1vb5t83WQAb9AtSdImzpGzgpsyc0p7MGuzYtWK9mDWZunypUyZOaUnS5MkSVVgOCu4hUsWVmVeSZJUTIazgqsbUFeVeSVJUjEZzgpu2vhp1PavXaNty35b0r9f/zXaavvXMm38tJ4sTZIkVYHhrOAmjJ7AjNNmMHDAQIJg4ICBXD3xaq6aeNUabTNOm+HFAJIk9QGRmb1dQ7dobGzM1tbW3i5DkiRpvSJibmY2djbNkTNJkqQCMZxJkiQViOFMkiSpQAxnkiRJBWI4kyRJKhDDmSRJUoEYziRJkgrEcCZJklQghjNJkqQCMZxJkiQViOFMkiSpQAxnkiRJBWI4kyRJKhDDmSRJUoEYziRJkgqkquEsIo6NiIcj4rGIOG8d850YERkRjeXXR0XE3Ij4XfnfI6pZpyRJUlFsUa0VR0Q/YDpwFPAUMCciZmXmQx3m2x74JHBvRfNzwHsz85mIaABuA/asVq2SJElFUc2Rs1HAY5k5PzOXAzcCJ3Qy30XApcCytobM/L/MfKb88kFgm4jYqoq1SpIkFUI1w9mewJMVr5+iw+hXRLwN2Dszf7iO9ZwI/CYzX+v+EiVJkoqlaoc11yciaoCvAhPXMc+BlEbVjl7L9MnAZIC6urruL1KSJKmHVXPk7Glg74rXe5Xb2mwPNAB3RsQTwGhgVsVFAXsBM4EPZ+bjnW0gM2dkZmNmNu6yyy5V6IIkSVLPqmY4mwMMjohBEdEfOAmY1TYxM1/KzJ0zsz4z64HZwLjMbI2IHYEfAudl5q+qWKMkSVKhVC2cZeZK4CxKV1r+HvhOZj4YERdGxLj1LH4W8PfA+RFxf/nx5mrVKkmSVBSRmb1dQ7dobGzM1tbW3i5DkiRpvSJibmY2djbNOwRIkiQViOFMkiSpQAxnkiRJBWI4kyRJKhDDmSRJUoEYziRJkgrEcCZJklQghjNJkqQCMZxJkiQViOFMkiSpQAxnkiRJBWI4kyRJKhDDmSRJUoEYziRJkgrEcCZJklQghjNJkqQCMZxJkiQViOFMkiSpQAxnkiRJBWI4kyRJKhDDmSRJUoEYziRJkgrEcCZJklQghjNJkqQCMZxJkiQViOFMkiSpQAxnkiRJBWI4kyRJKhDDmSRJUoEYziRJkgrEcCZJklQghjNJkqQCMZxJkiQViOFMkiSpQAxnkiRJBVLVcBYRx0bEwxHxWESct475ToyIjIjGirbPl5d7OCKOqWadkiRJRbFFtVYcEf2A6cBRwFPAnIiYlZkPdZhve+CTwL0VbUOAk4ADgT2A2yPiLZm5qlr1SpIkFUE1R85GAY9l5vzMXA7cCJzQyXwXAZcCyyraTgBuzMzXMvOPwGPl9UmSJPVp1QxnewJPVrx+qtzWLiLeBuydmT/c0GUlSZL6ol67ICAiaoCvAp/diHVMjojWiGhdvHhx9xUnSZLUS6oZzp4G9q54vVe5rc32QANwZ0Q8AYwGZpUvCljfsgBk5ozMbMzMxl122aWby5ckSep51Qxnc4DBETEoIvpTOsF/VtvEzHwpM3fOzPrMrAdmA+Mys7U830kRsVVEDAIGA/dVsVZJkqRCqNrVmpm5MiLOAm4D+gFXZeaDEXEh0JqZs9ax7IMR8R3gIWAl0OSVmpIkaXMQmdnbNXSLxsbGbG1t7e0yJEmS1isi5mZmY2fTvEOAJElSgRjOJEmSCsRwJkmSVCCGM0mSpAIxnEmSJBWI4UySJKlADGeSJEkFYjiTJEkqEMOZJElSgRjOJEmSCsRwJkmSVCCGM0mSpAIxnEmSJBWI4UySJKlADGeSJEkFYjiTJEkqEMOZJElSgRjOJEmSCsRwJkmSVCCGM0mSpAIxnEmSJBWI4UySJKlADGeSJEkFYjiTJEkqEMOZJElSgRjOJEmSCsRwJkmSVCCGM0mSpAIxnEmSJBWI4UySJKlADGeSJEkFYjiTJEkqEMOZJElSgRjOJEmSCqSq4Swijo2IhyPisYg4r5PpH4uI30XE/RFxd0QMKbdvGRHXlqf9PiI+X806JUmSiqJq4Swi+gHTgXcDQ4CT28JXheszc2hmjgC+Any13P4BYKvMHAocBHw0IuqrVWtXtMxuof7cemom1VB/bj0ts1t6sxxJktRHbVHFdY8CHsvM+QARcSNwAvBQ2wyZ+ZeK+bcFsm0SsG1EbAFsAywHKuftUS2zW5h83WSWLl8KwIIlC5h83WQAJoye0FtlSZKkPqiahzX3BJ6seP1UuW0NEeHReG8AAB0LSURBVNEUEY9TGjk7u9x8C/AqsAhYCFyemUuqWOs6TZk5pT2YtVm6fClTZk7ppYokSVJf1esXBGTm9MzcFzgX+Ndy8yhgFbAHMAj4bETs03HZiJgcEa0R0bp48eKq1bhwycINapckSXqjqhnOngb2rni9V7ltbW4E3ld+fgrwk8xckZnPAr8CGjsukJkzMrMxMxt32WWXbir79eoG1G1QuyRJ0htVzXA2BxgcEYMioj9wEjCrcoaIGFzx8jjg0fLzhcAR5Xm2BUYDf6hires0bfw0avvXrtFW27+WaeOn9VJFkiSpr6raBQGZuTIizgJuA/oBV2XmgxFxIdCambOAsyLiSGAF8AJwennx6cDVEfEgEMDVmflAtWpdn7aT/qfMnMLCJQupG1DHtPHTvBhAkiR1u8jM9c+1CWhsbMzW1tbeLkOSJGm9ImJuZr7ulC0owAUBkiRJ+hvDmSRJUoEYziRJkgrEcCZJklQghjNJkqQCMZxJkiQViOFMkiSpQAxnkiRJBWI4kyRJKhDDmSRJUoH0mds3RcRiYEGVVr8z8FyV1r0psP/23/5vvjb3/oPvgf2vTv8HZuYunU3oM+GsmiKidW33v9oc2H/7b//tf2/X0Zs29/fA/vd8/z2sKUmSVCCGM0mSpAIxnHXNjN4uoJfZ/82b/d+8be79B98D+9/DPOdMkiSpQBw5kyRJKpDNMpxFxLER8XBEPBYR53UyfauIuKk8/d6IqK+Y9vly+8MRcUxX11kkb7T/EXFURMyNiN+V/z2iYpk7y+u8v/x4c8/1aMNsRP/rI+KvFX38ZsUyB5Xfl8ci4sqIiJ7r0YbZiP5PqOj7/RGxOiJGlKf1pf3/zoj4TUSsjIj3d5h2ekQ8Wn6cXtHel/Z/p/2PiBERcU9EPBgRD0TEhyqmXRMRf6zY/yN6qj8baiP3/6qKPs6qaB9U/qw8Vv7s9O+JvrwRG7H/x3b4/C+LiPeVp/Wl/f+ZiHio/DP+s4gYWDGt5z7/mblZPYB+wOPAPkB/4LfAkA7zfAL4Zvn5ScBN5edDyvNvBQwqr6dfV9ZZlMdG9v+twB7l5w3A0xXL3Ak09nb/qtz/emDeWtZ7HzAaCODHwLt7u6/d3f8O8wwFHu+j+78eGAb8D/D+ivYBwPzyvzuVn+/UB/f/2vr/FmBw+fkewCJgx/LrayrnLepjY/pfnvbKWtb7HeCk8vNvAh/v7b5Wo/8V8wwAlgC1fXD/j63o18f52+//Hv38b44jZ6OAxzJzfmYuB24ETugwzwnAteXntwDvKifhE4AbM/O1zPwj8Fh5fV1ZZ1G84f5n5v9l5jPl9geBbSJiqx6puvtszP7vVETsDrwpM2dn6ZP6P8D7ur/0btFd/T+5vOymZr39z8wnMvMBYHWHZY8B/jczl2TmC8D/Asf2tf2/tv5n5iOZ+Wj5+TPAs0CnX6BZYBuz/ztV/mwcQemzAqXPTp/b/x28H/hxZi6tXqlV0ZX+31HRr9nAXuXnPfr53xzD2Z7AkxWvnyq3dTpPZq4EXgL+bh3LdmWdRbEx/a90IvCbzHytou3q8pD2Fwp8WGdj+z8oIv4vIn4REYdWzP/UetZZFN21/z8E3NChra/s/w1dtq/t//WKiFGURh4er2ieVj4U9O8F/k/bxvZ/64hojYjZbYf0KH02Xix/Vt7IOntSd/2tOonXf/774v7/R0ojYetatiqf/80xnGkjRcSBwKXARyuaJ2TmUODQ8uO03qityhYBdZn5VuAzwPUR8aZerqnHRcTbgaWZOa+ieXPY/6J9pPg64IzMbBtd+TywPzCS0mGfc3upvGobmKVvij8FuCIi9u3tgnpaef8PBW6raO5z+z8iTgUagct6Y/ubYzh7Gti74vVe5bZO54mILYAdgOfXsWxX1lkUG9N/ImIvYCbw4cxs/19zZj5d/vdl4HpKw8dF9Ib7Xz6c/TxAZs6lNGrwlvL8e1Us32f3f9nr/tfcx/b/hi7b1/b/WpX/M/JDYEpmzm5rz8xFWfIacDV9c/9X/pzPp3Se5VspfTZ2LH9WNnidPaw7/lZ9EJiZmSvaGvra/o+II4EpwLiKo0M9+vnfHMPZHGBw+eqa/pT+0MzqMM8soO1KjPcDPy8fS54FnBSlq9kGAYMpnQjYlXUWxRvuf0TsSOkX83mZ+au2mSNii4jYufx8S+B4YB7FtDH93yUi+gFExD6U9v/8zFwE/CUiRpcP530YuLUnOvMGbMzPPxFRQ+mXc/v5Zn1w/6/NbcDREbFTROwEHA3c1gf3f6fK888E/iczb+kwbffyv0HpfJs+t//L+32r8vOdgUOAh8qfjTsofVag9Nnpc/u/wsl0+M9ZX9r/EfFW4FuUgtmzFZN69vO/sVcUbIoP4D3AI5RGPqaU2y4s7wyArYGbKZ3wfx+wT8WyU8rLPUzFFRmdrbOojzfaf+BfgVeB+ysebwa2BeYCD1C6UOBrQL/e7mcV+n9iuX/3A78B3luxzkZKv5AeB/6T8hc8F/GxkT//hwOzO6yvr+3/kZTOG3mV0qjIgxXLfqT8vjxG6bBeX9z/nfYfOBVY0eHzP6I87efA78rvwbeB7Xq7n1Xo/8HlPv62/O8/Vqxzn/Jn5bHyZ2er3u5nd/e/PK2e0qhQTYd19qX9fzvw54qf8VkVy/bY5987BEiSJBXI5nhYU5IkqbAMZ5IkSQViOJMkSSoQw5kkSVKBGM4kSZIKZIv1zyJJm46YFDXAPwPDgVcofYnuCuDKbM57K+bbkdINm98LTM3mvKjc/hbgKkpfI3B6NueLHda/LTADOCSbs77a/ZG0+XHkTFJfMx14Uzbnydmck7I5P0jpe41OrpypHLpOonTv0NkV7Y8A84GTOgaz8vRXKX3foSRVhSNnkvqMmBT1lO7r+eYOk64A3tZx/mzOZTEprqd0g+P/La9jN0qBbUVMiu9T+sLprYHnszkv6GSbHwa+DEyg9A3k3wZ2yuY8vDx9LKVvjV9E6Us8LygHQEnqlCNnkvqSkcDCbM6llY3ZnC9lc96xlmX+G3hfTIq/K78+Hbi2/PyabM7PZXOeBTTGpHh7x4WzOf8HeLT8/FVKd0gAoLzOq4GmbM7Pl7f1X2+4d5I2C4YzSX3JBt/yJJvz/4CHKN2eCOCgbM5WYBVQF5Piv2NSXAEMonSj+w0xBqgF/i0mxTcpja7129AaJW1ePKwpqS+ZA+wdk6K2cvQsJsWbgF2zOR9dy3L/DXw0JsVc4O5y20mUbmLcmM25KibFNaw9WCV/+8/ulh2mPZ/N+bGKWrbbkA5J2vw4ciapz8jmXEDpCszzOkz6GnDkOhZtAQYDl1I6Zwzg74CXsjlXlV/XrWP5PwG7l5+PqGi/B3hz+Vy4tvPZblpXHyTJG59L6lPKX6XxCeBgSif2DwDuyea8Yj3LtQBbZHN+qPx6B+Dm8joWAO+i9PUanwU+A4wHvpTNeUlMijHAV4FfA8spjbhNzeZsjknxLuCTwGPATsBF2Zzzu7fXkvoSw5kkSVKBeFhTkiSpQAxnkiRJBWI4kyRJKhDDmSRJUoEYziRJkgrEcCZJklQghjNJkqQCMZxJkiQViPfWVOHMnTt3r5qamp+uXr16fyB6ux5JKoisqan5w+rVq48+6KCDnurtYlQ9hjMVTk1NzU933XXXwbvttlvU1Di4K0kAq1evjkWLFu23YMGC+8aNG/fuWbNm/ba3a1J1+JdPhbN69er9d9ttty0MZpL0NzU1Ney+++41/fv33x34zLhx497W2zWpOvzrpyJyxEySOlFTU0NEAPwFeEcvl6Mq8S+g1Ik//elPnHTSSey7774cdNBBvOc97+GRRx7Z6PVOnDiRW2655XXtra2tnH322Ru9foBrrrmGs846q0vzXnHFFWy99da89NJL3bLtnlJfX8/QoUMZMWIEI0aM4Ne//jUAxx57LDvuuCPHH3981ba9tn04ceJEBg0a1F7TwQcf3O3bnjp1KpdffjkA559/PrfffjtQ2o9Lly7tlm3cfPPNHHDAAYwdO/Z10x588EGOOOII9ttvPwYPHsxFF11EZnbLdtdlY/v3zDPP8P73vx+A+++/nx/96EcbtPyf//xnjj/+eIYPH86QIUN4z3ve84Zr6UYrga17uwhVh+FM6iAzGT9+PIcffjiPP/44c+fO5ctf/jJ//vOfq7bNxsZGrrzyyqqtf21uuOEGRo4cyXe/+91uWd+qVau6ZT1dcccdd3D//fdz//33twehc845h+uuu67Haujosssua6+pLTBWy4UXXsiRRx4JdG84++///m+am5u544471mj/61//yrhx4zjvvPN4+OGH+e1vf8uvf/1rvv71r3fLdtdm1apVG92/PfbYoz1Qv5Fwdv7553PUUUfx29/+loceeohLLrnkDdfSZuXKlRu9DvVdhjNt8lpmt1B/bj01k2qoP7eeltktG7W+O+64gy233JKPfexj7W3Dhw/n0EMPJTM555xzaGhoYOjQodx0000A3HnnnRx22GGccMIJ7LPPPpx33nm0tLQwatQohg4dyuOPP96+rttvv53Gxkbe8pa38IMf/KB9+bbRnqlTp/KRj3yEww8/nH322WeN0Pbtb3+bUaNGMWLECD760Y+2h6Grr76at7zlLYwaNYpf/epXXern448/ziuvvMLFF1/MDTfcAMA3v/lNzjnnnPZ5Kkfh1rbt7bbbjs9+9rMMHz6ce+65hwsvvJCRI0fS0NDA5MmT20dW5syZw7BhwxgxYkT7ewilP77nnHMOI0eOZNiwYXzrW9/q6q56nXe9611sv/3265ynubmZkSNHMnz4cE488cT2P/oTJ07k7LPP5uCDD2afffZp/2OemZx11lnst99+HHnkkTz77LMbVNPzzz/P0UcfzYEHHsiZZ57JwIEDee6553jiiSfa3wOAyy+/nKlTp66zxkptI3hXXnklzzzzDGPHjmXs2LFcddVVfOpTn1qjv5/+9Kdft/wNN9zA0KFDaWho4NxzzwVKge/uu+/mH//xH9f4OQC4/vrrOeSQQzj66KMBqK2t5T//8z/bg8rUqVM57bTTGDNmDIMHD6a5uRko/Wy/853v5LjjjmO//fbjYx/7GKtXr15rDbDmz9S0adPW6F/b9Da33HILEydObH9POtuHbe/18uXLOf/887npppsYMWIEN910E4MHD2bx4sUArF69mr//+79vf91m0aJF7LXXXu2vhw0b1v780ksvZejQoQwfPpzzzjsPKAXA0aNHM2zYMMaPH88LL7wAwOGHH86nPvUpGhsb+drXvsbcuXM57LDDOOiggzjmmGNYtGgRAFdeeSVDhgxh2LBhnHTSSa/bd9oMZKYPH4V6tLa2Zld9+55vZ+0napMzaX/UfqI2v33Pt7u8jo6+9rWv5ac+9alOp91yyy155JFH5sqVK/NPf/pT7r333vnMM8/kHXfckTvssEM+88wzuWzZstxjjz3y/PPPz8zMK664Ij/5yU9mZubpp5+exxxzTK5atSofeeSR3HPPPfOvf/1r3nHHHXncccdlZuYFF1yQY8aMyWXLluXixYtzwIABuXz58nzooYfy+OOPz+XLl2dm5sc//vG89tpr85lnnsm99947n3322Xzttdfy4IMPzqampszMvPXWW/MLX/hCp325+OKL88ILL8xVq1ZlXV1d/ulPf8pnn30299133/Z5jj322PzlL3+51m1nZgJ50003tS/z/PPPtz8/9dRTc9asWZmZeeCBB+avf/3rzMw899xz88ADD8zMzG9961t50UUXZWbmsmXL8qCDDsr58+evcx8NHDgwGxoacvjw4Tlq1Kg1plW+l5157rnn2p9PmTIlr7zyysws7Zv3v//9uWrVqnzwwQfb34f/9//+X/s+f/rpp3OHHXbIm2+++XXrPf3007O+vj6HDx+ew4cPz1NOOSUzM//pn/4pv/jFL2Zm5g9+8IMEcvHixfnHP/6x/T3IzLzsssvyggsuWGeNF1xwQV522WXt22urY+DAgbl48eLMzHz55Zdzn332ad9XY8aMyQceeGCNWp9++un2n5kVK1bk2LFjc+bMmZmZedhhh+WcOXNe179Pf/rTecUVV7yufccdd8yXXnopL7jgghw2bFguXbo0Fy9enHvttVc+/fTTeccdd+RWW22Vjz/+eK5cuTKPPPLIvPnmm9dZQ8efqcr+ZWZuu+227c9vvvnmPP3009vfk872YeV7ffXVV7d/PjIzp06dmv/+7/+emZm33XZb/sM//MPr+viTn/wkd9hhhzz88MPz4osvzqeffjozM3/0ox/lmDFj8tVXX83Mv/3sDx06NO+8887MzPzCF77Q/vk/7LDD8uMf/3hmZi5fvjzHjBmTzz77bGZm3njjjXnGGWdkZubuu++ey5Yty8zMF1544XX1tLa25nvf+95/f+973/vPWYDf2T66/+FXaWiTNmXmFJYuX3NUYenypUyZOYUJoyd0+/buvvtuTj75ZPr168euu+7KYYcdxpw5c3jTm97EyJEj2X333QHYd99920cYhg4dusYhog9+8IPU1NQwePBg9tlnH/7whz+8bjvHHXccW221FVtttRVvfvOb+fOf/8zPfvYz5s6dy8iRI4HSYaY3v/nN3HvvvRx++OHssssuAHzoQx9qPz9u3LhxjBs3rtO+3HDDDcycOZOamhpOPPFEbr75Zs466yz22WcfZs+ezeDBg/nDH/7AIYccwvTp0zvdNkC/fv048cQT29d7xx138JWvfIWlS5eyZMkSDjzwQA499FBefvllxowZA8App5zSPmr405/+lAceeKB9lOOll17i0UcfZdCgQevcF3fccQc777zzOufpzLx58/jXf/1XXnzxRV555RWOOeaY9mnve9/7qKmpYciQIe2Hse+66672fb7HHntwxBFHrHXdl112Wfu5TW3uuuuu9sPGxx13HDvttNNG1bg+2223HUcccQQ/+MEPOOCAA1ixYgVDhw5dY545c+as8TMzYcIE7rrrLt73vvd1eTudOeGEE9hmm23YZpttGDt2LPfddx877rgjo0aNYp999gHg5JNP5u6772bLLbdcaw0df6Y2RGf7cF0+8pGPcMIJJ/CpT32Kq666ijPOOON18xxzzDHMnz+fn/zkJ/z4xz/mrW99K/PmzeP222/njDPOoLa2FoABAwbw0ksv8eKLL3LYYYcBcPrpp/OBD3ygfV0f+tCHAHj44YeZN28eRx11FFAaQW77/TFs2DAmTJjA+973vo3eJ9o0Gc60SVu4ZOEGtXfFgQce2OkJ3+uz1VZbtT+vqalpf11TU7PG+SXlK63W+rrjuvr168fKlSvJTE4//XS+/OUvrzHv9773vQ2u9Xe/+x2PPvpo+x+G5cuXM2jQIM466yxOOukkvvOd77D//vszfvx4ImKt2wbYeuut6devHwDLli3jE5/4BK2trey9995MnTqVZcuWrbOWzOQ//uM/NiiAbIyJEyfyve99j+HDh3PNNddw5513tk+rfN8zq3ui+xZbbNF+eA9Y431aV41dceaZZ/KlL32J/fffv9Ow8UYMGTKEu+66a422+fPns9122/GmN70JWPvPdld+5itV/kx1pnL5jj9fG7oP9957b3bddVd+/vOfc99999HS0vlpEQMGDOCUU07hlFNO4fjjj3/de9FV2267bXttBx54IPfcc8/r5vnhD3/IXXfdxfe//32mTZvG7373O7bYwj/XmxPPOdMmrW5A3Qa1d8URRxzBa6+9xowZM9rbHnjgAX75y19y6KGHctNNN7Fq1SoWL17MXXfdxahRozZo/TfffDOrV6/m8ccfZ/78+ey3335dWu5d73oXt9xyS/s5T0uWLGHBggW8/e1v5xe/+AXPP/88K1as4Oabb17vum644QamTp3KE088wRNPPMEzzzzDM888w4IFCxg/fjy33norN9xwQ/v5Lmvbdkdtfyh33nlnXnnllfaQu+OOO7L99ttz7733AnDjjTe2L3PMMcfwjW98gxUrVgDwyCOP8OqrrwKw//77d+m92RAvv/wyu+++OytWrFjrH+JK73znO9v3+aJFi153onxXlr/++usB+PGPf9x+/tGuu+7Ks88+y/PPP89rr73WPpL4Rmrcfvvtefnll9tfv/3tb+fJJ5/k+uuv5+STT37d/KNGjeIXv/gFzz33HKtWreKGG25oH+lZmwkTJnD33Xe3XyH617/+lbPPPpt//ud/bp/n1ltvZdmyZTz//PPceeed7SOt9913H3/84x9ZvXo1N910E+94xzs2qIaO/dt11135/e9/z+rVq5k5c+Z63591rQtKYfbUU0/lAx/4QKeh8Oc//3n7eX8vv/wyjz/+OHV1dRx11FFcffXV7dOWLFnCDjvswE477cQvf/lLAK677rpO+7XffvuxePHi9nC2YsUKHnzwQVavXs2TTz7J2LFjufTSS3nppZd45ZVXNqiP2vQZzrRJmzZ+GrX9a9doq+1fy7Tx097wOiOCmTNncvvtt7Pvvvty4IEH8vnPf57ddtuN8ePHM2zYMIYPH84RRxzBV77yFXbbbbcNWn9dXR2jRo3i3e9+N9/85jfZeuuuXQ0/ZMgQLr74Yo4++miGDRvGUUcdxaJFi9h9992ZOnUqY8aM4ZBDDuGAAw5oX2bWrFmcf/75r1vXjTfeyPjx49doGz9+PDfeeCM77bQTBxxwAAsWLGgPnmvbdkc77rgjkyZNoqGhgWOOOab9jzOUrgKcNGkSI0aM4NVXX2WHHXYASn8YhwwZwtve9jYaGhr46Ec/ysqVK3nuuec2ePTq0EMP5QMf+AA/+9nP2GuvvbjtttteN89FF13E29/+dg455JAuhb/x48czePBghgwZwoc//OH2Q7OdOeecc9q/SmPEiBEsX76cCy64gLvuuosDDzyQ7373u9TVlf7jsOWWW3L++eczatQojjrqqDVq2dAaJ0+ezLHHHrvG11988IMf5JBDDun0MOruu+/OJZdcwtixYxk+fDgHHXQQJ5xwwjq3sc0223Drrbdy8cUXs99++zF06FBGjhy5xte2DBs2jLFjxzJ69Gi+8IUvsMceewC0z3fAAQcwaNAgxo8fv0E1dOzfJZdcwvHHH8/BBx/cfiiwq8aOHctDDz3UfkEAlA7/v/LKK2sdZZw7dy6NjY0MGzaMMWPGcOaZZzJy5EiOPfZYxo0bR2NjIyNGjGj/mpNrr72Wc845h2HDhnH//fd3+hns378/t9xyC+eeey7Dhw9v/0qYVatWceqppzJ06FDe+ta3cvbZZ7PjjjtuUB+16YtqD91LG2ru3Ll50EEHdXn+ltktTJk5hYVLFlI3oI5p46dV5XwzbZxXXnml/Sq7Sy65hEWLFvG1r31trfP/4Ac/YP78+d32/W9FUV9fT2tr6xs6X25DHH/88Xz605/mXe96V1W302bq1Klst912fO5zn1uj/c477+Tyyy9fY2SwaFpbW/n0pz/dPtpVdHPnzuWLX/ziFcCiWbNmfaW361H38yC2NnkTRk8wjG0CfvjDH/LlL3+ZlStXMnDgQK655pp1zl/NL5Lty1588UVGjRrF8OHDeyyYbcouueQSvvGNb3Tp8LHUUxw5U+Fs6MiZJG1OHDnr+zznTJIkqUAMZyqirPyKAUlSyerVq6v+NS/qfYYzFU5NTc0fFi1atNqAJkl/s3r1ahYtWrR62bJlzwHr/rI4bdK8IECFs3r16qOf+P/t3bEJwlAUBdA7gGAnjiFkjz+OC/0RnMMdUggBC1sDFrGw1lQhX3LOBK+8zX2376/DMBzmnlUCbMU0TRnH8VFrrUn2SeYnEPhLwhnN6bruVko5JTkn+b1iDbA9+yT3JJe1D2EZ2po0q5SyS3JM8n3HBWB7Xvk0NZ9rH8IyhDMAgIYoBAAANEQ4AwBoiHAGANCQNzQ68s+9VyfqAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fairf1combined(X_train_g, y_train_g, g_train, np.arange(0.001, 0.2, 0.002))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "uq7trdAEJ2K3"
   },
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "qwf9Z0EsHF1W"
   },
   "source": [
    "## (6b) Fairness-based Machine Learning Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "LNwf4UiKJpyO"
   },
   "source": [
    "### Functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {
    "id": "rkQvmDdSFYxf"
   },
   "outputs": [],
   "source": [
    "#need to find a sweetspot between the fairness and accuracy\n",
    "#we want high accuracy and low equal opp \n",
    "#we would then want high 1-(low equal opp)\n",
    "#add (accuracy and high 1-(low equal opp)) - 1 would be the best combined score\n",
    "\n",
    "#looking into the f1 score \n",
    "\n",
    "def rw_fairaccuracyf1combined(X_train, y_train, train, cvals):\n",
    "  #empty lists for our scores\n",
    "  fairness = []\n",
    "  equ_opp = []\n",
    "  f1score = []\n",
    "  #mitgating the bias\n",
    "  #assigning the new weights to the unprivileged and privileged groups, transform the dataset with reweighing\n",
    "  RW = Reweighing(unprivileged_groups=unprivileged_groups,\n",
    "                privileged_groups=privileged_groups)\n",
    "  #obtaining a set of weights for the training set\n",
    "  train = RW.fit_transform(train)\n",
    "\n",
    "\n",
    "#looping through each C value\n",
    "  for i in cvals:\n",
    "    #our model\n",
    "    clf = LogisticRegression(solver='liblinear', C=i, random_state=1)\n",
    "    #accuracy score\n",
    "    score = cross_val_score(clf, X_train, y_train, fit_params={'sample_weight': train.instance_weights}, cv = 5)\n",
    "    #prediction labels for fairness\n",
    "    fair_pred = cross_val_predict(clf, X_train, y_train, fit_params={'sample_weight': train.instance_weights}, cv = 5)\n",
    "    test_pred = train.copy()\n",
    "    test_pred.labels = fair_pred\n",
    "\n",
    "    #defining our metric\n",
    "    metric = ClassificationMetric(train, test_pred, unprivileged_groups=unprivileged_groups, privileged_groups=privileged_groups)\n",
    "\n",
    "    #recording the scores\n",
    "    equ_opp.append(metric.equal_opportunity_difference())\n",
    "    f1score.append(f1_score(y_train, fair_pred, average = 'macro'))\n",
    "\n",
    "  #taking the absolute value for the Equality of Opportunity scores\n",
    "  absol_equ_opp = []\n",
    "  for value in equ_opp:\n",
    "    absolute = abs(value)\n",
    "    absol_equ_opp.append(absolute)\n",
    "\n",
    "  #combining the values of accuracy and Equality of Opportunity\n",
    "  combined = [(1-j) + k for (j, k) in zip(absol_equ_opp, f1score)]\n",
    "\n",
    "\n",
    "  #re-scaling the combined value to be out of 1, closer to 1 is better\n",
    "  rescaledcombined = []\n",
    "  for value in combined:\n",
    "    rescaled = value - 1\n",
    "    rescaledcombined.append(rescaled)\n",
    "  \n",
    "  best_c = equ_opp.index(min(equ_opp, key= abs))\n",
    "  best_combined = rescaledcombined.index(max(rescaledcombined)) # we want this to be \n",
    "  best_f1 = f1score.index(max(f1score)) #we want f1 score to be 1\n",
    "\n",
    "  print('Best Equality of Opportunity score is at C: ', cvals[best_c], 'with an equality of opportunity score of', min(equ_opp, key= abs))\n",
    "  print('Best f1 score is at C: ', cvals[best_f1], 'with a score of', max(f1score))\n",
    "  print('Best combined score is at C: ', cvals[best_combined], 'with a score of', max(rescaledcombined))\n",
    "\n",
    "  #pltting the results\n",
    "  xplt = cvals\n",
    "  plt.figure(figsize = (10,6))\n",
    "  plt.scatter(xplt, rescaledcombined, color = 'magenta', label = 'Combined: Average, F1 and Equality of Opportunity Scores')\n",
    "  plt.legend(loc='upper center', bbox_to_anchor=(0.5, -0.10),\n",
    "          fancybox=True, shadow=True)\n",
    "\n",
    "  font1 = {'family':'serif','color':'magenta','size':13}\n",
    "  font2 = {'family':'serif','color':'magenta','size':11}\n",
    "  plt.title('Combined Scores', fontdict = font1, color = 'magenta')\n",
    "  plt.xlabel('C Value', fontdict = font2)\n",
    "  plt.ylabel('Score', fontdict = font2)\n",
    "  plt.show()\n",
    "\n",
    "  #return(df) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "SBM9O6c-JzLN"
   },
   "source": [
    "\n",
    "### Adult"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 481
    },
    "id": "k17odTooJQSq",
    "outputId": "40cd382f-a72f-4399-efd7-3e640fca17b3"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best Equality of Opportunity score is at C:  0.01 with an equality of opportunity score of -0.011180795108922947\n",
      "Best f1 score is at C:  0.005 with a score of 0.6794571444787936\n",
      "Best combined score is at C:  0.01 with a score of 0.6663937880442259\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#calculating the combined score on the reweighed with a broad c\n",
    "\n",
    "\n",
    "rw_fairaccuracyf1combined(X_train_a, y_train_a, a_train, c)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 481
    },
    "id": "HLEL-JSmKkkv",
    "outputId": "e9f862e9-b423-4170-e8fe-f09ef65e3a90"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best Equality of Opportunity score is at C:  0.009000000000000001 with an equality of opportunity score of -0.011180795108922947\n",
      "Best f1 score is at C:  0.005 with a score of 0.6794571444787936\n",
      "Best combined score is at C:  0.009000000000000001 with a score of 0.6663937880442259\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#narrowing in to more specific cs\n",
    "\n",
    "rw_fairaccuracyf1combined(X_train_a, y_train_a, a_train, np.arange(0.005, 0.03, 0.001))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "jJQLV8zJJ3CO"
   },
   "source": [
    "### German"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 481
    },
    "id": "nXWPnbMCJ9Il",
    "outputId": "67ae9ff5-0c47-4c3a-b638-c1cc85de3df1"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best Equality of Opportunity score is at C:  0.01 with an equality of opportunity score of 0.0031311154598824276\n",
      "Best f1 score is at C:  0.005 with a score of 0.5257516199573378\n",
      "Best combined score is at C:  0.005 with a score of 0.518413068098238\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#calculating the combined score on the reweighed with a broad c\n",
    "\n",
    "rw_fairaccuracyf1combined(X_train_g, y_train_g, g_train, c)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 481
    },
    "id": "2R4LClrtLaFC",
    "outputId": "dbeb643f-307b-4dd8-bedd-62ac0778961f"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best Equality of Opportunity score is at C:  0.009000000000000001 with an equality of opportunity score of 0.0031311154598824276\n",
      "Best f1 score is at C:  0.002 with a score of 0.5726776674774834\n",
      "Best combined score is at C:  0.002 with a score of 0.540583734013687\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#narrowing in to more specific cs\n",
    "\n",
    "rw_fairaccuracyf1combined(X_train_g, y_train_g, g_train, np.arange(0.001, 0.02, 0.001))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "mi_RjLM3J3Gs"
   },
   "source": [
    "## Task 3b: Compute and Report the Final Scores"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "FCxZGsETLou_"
   },
   "source": [
    "### Adult"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "7LAf4xU_MyLY"
   },
   "source": [
    "<font color='darkgreen'> **(5a) Standard Logistic Regression Model**</font>\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "q4prBPs1LuOY",
    "outputId": "0d0aee85-e483-4dd1-95ca-1693967ccf85"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The final Accuracy score on the test data with hyperparameter C: 0.001 is 0.8042721626970586\n",
      "The final Equality of Opportunity score at C: 0.001 is [-0.4511051574012056]\n",
      "The disparate impact score is [0.0]\n"
     ]
    }
   ],
   "source": [
    "#calculating the final scores\n",
    "\n",
    "finalscores(X_train_a, y_train_a, X_test_a, y_test_a, a_test, 0.001)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "oOd4gOiNUP0t"
   },
   "source": [
    "> The <font color='darkgreen'>**final *accuracy* score**</font> in testing data at hyperparameter C: 0.001 is <font color='darkgreen'>**0.8043**</font> and the <font color='darkgreen'>**final *Equality of Opportunity* score**</font> is <font color='darkgreen'>**-0.4511**</font>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "BzP86wXdM2IU"
   },
   "source": [
    "<font color='magenta'> **(6b) Fairness-based Machine Learning Model** </font>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "jteU1dgFM59j",
    "outputId": "2ea883e7-2519-4bd7-fd8b-cc816e23a5a8"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "After reweighing, the final accuracy score on the test data with hyperparameter C: 0.009 is 0.7909643076503106\n",
      "The Equality of Opportunity score at C: 0.009 is [0.02028996011236517]\n",
      "The disparate impact score is [0.6827554035209011]\n"
     ]
    }
   ],
   "source": [
    "#caluclating the final scores for the reweighed\n",
    "\n",
    "rw_finalscores(X_train_a, y_train_a, X_test_a, y_test_a,a_train, a_test, 0.009)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "o339WaGkU8Im"
   },
   "source": [
    "> The <font color='magenta'>**final *accuracy* score**</font> in testing data at hyperparameter C: 0.009 is <font color='magenta'>**0.79096**</font> and the <font color='magenta'>**final *Equality of Opportunity* score**</font> is <font color='magenta'>**0.02029**</font>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "tNIrrpE9LqmK"
   },
   "source": [
    "### German"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "m3WPj0imUFel"
   },
   "source": [
    "<font color='darkgreen'> **(5a) Standard Logistic Regression Model**</font>\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "OTpseODBLrjV",
    "outputId": "477a7f69-98c6-46a4-e54d-aae757f23452"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The final Accuracy score on the test data with hyperparameter C: 0.003 is 0.6566666666666666\n",
      "The final Equality of Opportunity score at C: 0.003 is [-0.1355517494494739]\n",
      "The disparate impact score is [0.8047817047817047]\n"
     ]
    }
   ],
   "source": [
    "#calculating the final scores\n",
    "\n",
    "finalscores(X_train_g, y_train_g, X_test_g, y_test_g, g_test, 0.003)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "f4OYoOZVVhyw"
   },
   "source": [
    "> The <font color='darkgreen'>**final *accuracy* score**</font> in testing data at hyperparameter C: 0.003 is <font color='darkgreen'>**0.6567**</font> and the <font color='darkgreen'>**final *Equality of Opportunity* score**</font> is <font color='darkgreen'>**-0.1355**</font>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "ftu8uiVEUH7c"
   },
   "source": [
    "<font color='magenta'> **(6b) Fairness-based Machine Learning Model** </font>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "5jtdall5MFfE",
    "outputId": "f06b3c92-72de-4d0f-fca2-b272c6e3c71f"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "After reweighing, the final accuracy score on the test data with hyperparameter C: 0.002 is 0.66\n",
      "The Equality of Opportunity score at C: 0.002 is [-0.06398336187912901]\n",
      "The disparate impact score is [0.869017094017094]\n"
     ]
    }
   ],
   "source": [
    "#calculating the final scores for the reweighed\n",
    "\n",
    "rw_finalscores(X_train_g, y_train_g, X_test_g, y_test_g,g_train, g_test, 0.002)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "GObzqgldVyjx"
   },
   "source": [
    "> The <font color='magenta'>**final *accuracy* score**</font> in testing data at hyperparameter C: 0.002 is <font color='magenta'>**0.66**</font> and the <font color='magenta'>**final *Equality of Opportunity* score**</font> is <font color='magenta'>**-0.06398**</font>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {
    "id": "5ogz03fqV5PD"
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "colab": {
   "collapsed_sections": [],
   "name": "244788_MachineLearningAssignment.ipynb",
   "provenance": [],
   "toc_visible": true
  },
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}