05_evaluation.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Part 5: Horse Race Prediction\n",
    "## Evaluation of Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import math\n",
    "\n",
    "from sklearn.model_selection import cross_val_score\n",
    "from sklearn.model_selection import KFold\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "\n",
    "from sklearn.linear_model import Ridge\n",
    "from sklearn.linear_model import Lasso\n",
    "\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.naive_bayes import GaussianNB\n",
    "from sklearn.ensemble import RandomForestClassifier, RandomForestRegressor\n",
    "from sklearn.svm import SVC\n",
    "\n",
    "from sklearn.metrics import mean_squared_error\n",
    "from sklearn.metrics import f1_score, roc_curve, auc, confusion_matrix, roc_auc_score\n",
    "from sklearn.metrics import precision_recall_curve\n",
    "from sklearn.metrics import plot_precision_recall_curve\n",
    "\n",
    "from sklearn.svm import SVR\n",
    "from sklearn.ensemble import GradientBoostingRegressor\n",
    "\n",
    "from imblearn.over_sampling import SMOTE\n",
    "from imblearn.pipeline import Pipeline\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "\n",
    "import time\n",
    "import joblib\n",
    "import pickle\n",
    "\n",
    "import warnings\n",
    "warnings.filterwarnings(\"ignore\")\n",
    "\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Read the train and test files\n",
    "df_train = pd.read_csv('./data/df_train.csv')\n",
    "df_test = pd.read_csv('./data/df_test.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(21139, 27)\n",
      "(5286, 27)\n"
     ]
    }
   ],
   "source": [
    "# View the shape of the train and test files\n",
    "print(df_train.shape)\n",
    "print(df_test.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<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>finishing_position</th>\n",
       "      <th>horse_number</th>\n",
       "      <th>horse_name</th>\n",
       "      <th>horse_id</th>\n",
       "      <th>jockey</th>\n",
       "      <th>trainer</th>\n",
       "      <th>actual_weight</th>\n",
       "      <th>declared_horse_weight</th>\n",
       "      <th>draw</th>\n",
       "      <th>length_behind_winner</th>\n",
       "      <th>...</th>\n",
       "      <th>running_position_6</th>\n",
       "      <th>race_id</th>\n",
       "      <th>recent_6_runs</th>\n",
       "      <th>recent_ave_rank</th>\n",
       "      <th>race_distance</th>\n",
       "      <th>HorseWin</th>\n",
       "      <th>HorseRankTop3</th>\n",
       "      <th>HorseRankTop50Percent</th>\n",
       "      <th>jockey_ave_rank</th>\n",
       "      <th>trainer_ave_rank</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>1.0</td>\n",
       "      <td>DOUBLE DRAGON</td>\n",
       "      <td>K019</td>\n",
       "      <td>B Prebble</td>\n",
       "      <td>D Cruz</td>\n",
       "      <td>133</td>\n",
       "      <td>1032</td>\n",
       "      <td>1</td>\n",
       "      <td>-</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2014-001</td>\n",
       "      <td>1</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1400</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>6.058185</td>\n",
       "      <td>7.369681</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>2.0</td>\n",
       "      <td>PLAIN BLUE BANNER</td>\n",
       "      <td>S070</td>\n",
       "      <td>D Whyte</td>\n",
       "      <td>D E Ferraris</td>\n",
       "      <td>133</td>\n",
       "      <td>1075</td>\n",
       "      <td>13</td>\n",
       "      <td>2</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2014-001</td>\n",
       "      <td>2</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1400</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>5.795734</td>\n",
       "      <td>6.721234</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>2 rows × 27 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   finishing_position  horse_number         horse_name horse_id     jockey  \\\n",
       "0                   1           1.0      DOUBLE DRAGON     K019  B Prebble   \n",
       "1                   2           2.0  PLAIN BLUE BANNER     S070    D Whyte   \n",
       "\n",
       "        trainer  actual_weight  declared_horse_weight  draw  \\\n",
       "0        D Cruz            133                   1032     1   \n",
       "1  D E Ferraris            133                   1075    13   \n",
       "\n",
       "  length_behind_winner  ...  running_position_6   race_id  recent_6_runs  \\\n",
       "0                    -  ...                 NaN  2014-001              1   \n",
       "1                    2  ...                 NaN  2014-001              2   \n",
       "\n",
       "   recent_ave_rank race_distance  HorseWin  HorseRankTop3  \\\n",
       "0              1.0          1400         1              1   \n",
       "1              2.0          1400         0              1   \n",
       "\n",
       "   HorseRankTop50Percent jockey_ave_rank trainer_ave_rank  \n",
       "0                      1        6.058185         7.369681  \n",
       "1                      1        5.795734         6.721234  \n",
       "\n",
       "[2 rows x 27 columns]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# View the first 2 rows of the train file\n",
    "df_train.head(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<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>finishing_position</th>\n",
       "      <th>horse_number</th>\n",
       "      <th>horse_name</th>\n",
       "      <th>horse_id</th>\n",
       "      <th>jockey</th>\n",
       "      <th>trainer</th>\n",
       "      <th>actual_weight</th>\n",
       "      <th>declared_horse_weight</th>\n",
       "      <th>draw</th>\n",
       "      <th>length_behind_winner</th>\n",
       "      <th>...</th>\n",
       "      <th>running_position_6</th>\n",
       "      <th>race_id</th>\n",
       "      <th>recent_6_runs</th>\n",
       "      <th>recent_ave_rank</th>\n",
       "      <th>race_distance</th>\n",
       "      <th>HorseWin</th>\n",
       "      <th>HorseRankTop3</th>\n",
       "      <th>HorseRankTop50Percent</th>\n",
       "      <th>jockey_ave_rank</th>\n",
       "      <th>trainer_ave_rank</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>11.0</td>\n",
       "      <td>SUPER FLUKE</td>\n",
       "      <td>T382</td>\n",
       "      <td>J Moreira</td>\n",
       "      <td>D Cruz</td>\n",
       "      <td>118</td>\n",
       "      <td>1118</td>\n",
       "      <td>6</td>\n",
       "      <td>-</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2016-138</td>\n",
       "      <td>1/5/3/2/4/5</td>\n",
       "      <td>4.941176</td>\n",
       "      <td>1650</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>4.070707</td>\n",
       "      <td>8.052941</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>10.0</td>\n",
       "      <td>PEARL WARM WARM</td>\n",
       "      <td>T183</td>\n",
       "      <td>S Clipperton</td>\n",
       "      <td>P O'Sullivan</td>\n",
       "      <td>118</td>\n",
       "      <td>1099</td>\n",
       "      <td>12</td>\n",
       "      <td>SH</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2016-138</td>\n",
       "      <td>2/3/8/7/1/6</td>\n",
       "      <td>5.266667</td>\n",
       "      <td>1650</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>6.714859</td>\n",
       "      <td>6.858209</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>2 rows × 27 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   finishing_position  horse_number       horse_name horse_id        jockey  \\\n",
       "0                   1          11.0      SUPER FLUKE     T382     J Moreira   \n",
       "1                   2          10.0  PEARL WARM WARM     T183  S Clipperton   \n",
       "\n",
       "        trainer  actual_weight  declared_horse_weight  draw  \\\n",
       "0        D Cruz            118                   1118     6   \n",
       "1  P O'Sullivan            118                   1099    12   \n",
       "\n",
       "  length_behind_winner  ...  running_position_6   race_id  recent_6_runs  \\\n",
       "0                    -  ...                 NaN  2016-138    1/5/3/2/4/5   \n",
       "1                   SH  ...                 NaN  2016-138    2/3/8/7/1/6   \n",
       "\n",
       "   recent_ave_rank race_distance  HorseWin  HorseRankTop3  \\\n",
       "0         4.941176          1650         1              1   \n",
       "1         5.266667          1650         0              1   \n",
       "\n",
       "   HorseRankTop50Percent jockey_ave_rank trainer_ave_rank  \n",
       "0                      1        4.070707         8.052941  \n",
       "1                      1        6.714859         6.858209  \n",
       "\n",
       "[2 rows x 27 columns]"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# View the first 2 rows of the test files\n",
    "df_test.head(2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Keep the features we want to train our model on\n",
    "X_train = df_train[['actual_weight', 'declared_horse_weight',\n",
    "                    'draw', #'win_odds', \n",
    "                    'jockey_ave_rank',\n",
    "                    'trainer_ave_rank', 'recent_ave_rank', 'race_distance']]\n",
    "\n",
    "y_train = df_train['HorseWin']\n",
    "\n",
    "# Keep the features we want to train our model on\n",
    "X_test = df_test[['actual_weight', 'declared_horse_weight',\n",
    "                   'draw', #'win_odds',\n",
    "                   'jockey_ave_rank',\n",
    "                    'trainer_ave_rank', 'recent_ave_rank', 'race_distance']]\n",
    "                    \n",
    "y_test = df_test[['HorseWin', 'HorseRankTop3']]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(21139, 7)"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Find the feature importances"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Open the pickled models\n",
    "smote_rfc = joblib.load('./model/smote_rfc_model.pkl')\n",
    "lgbm = joblib.load('./model/lgbm_model.pkl')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# find feature importance with random forest\n",
    "def feature_importance(model, X_train, y_train):\n",
    "    # fit the model\n",
    "    model.fit(X_train, y_train)\n",
    "\n",
    "    # get importance\n",
    "    importance = model[1][1].feature_importances_\n",
    "\n",
    "    # sort the index of the importance\n",
    "    sorted_idx = np.argsort(importance)\n",
    "\n",
    "    # name of the features\n",
    "    feature_names = X_train.columns[sorted_idx]\n",
    "\n",
    "    # summarize feature importance with feature names\n",
    "    for i,v in enumerate(importance[sorted_idx]):\n",
    "        print('Feature: %s, Score: %.5f' % (feature_names[i],v))\n",
    "\n",
    "    # plot a barh graph of feature importance\n",
    "    plt.figure(figsize=(10, 6))\n",
    "    plt.barh(feature_names, importance[sorted_idx])\n",
    "    plt.title('Feature Importance')\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Feature: race_distance, Score: 0.01995\n",
      "Feature: draw, Score: 0.02747\n",
      "Feature: actual_weight, Score: 0.03078\n",
      "Feature: declared_horse_weight, Score: 0.04338\n",
      "Feature: trainer_ave_rank, Score: 0.11216\n",
      "Feature: jockey_ave_rank, Score: 0.15601\n",
      "Feature: recent_ave_rank, Score: 0.61024\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAArYAAAF1CAYAAAAdl6lrAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjQuMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/MnkTPAAAACXBIWXMAAAsTAAALEwEAmpwYAAAo20lEQVR4nO3debidVX328e9NwiiQiAzGiMTaKKIoSgBRQbC8VqEWqSi0VEHsi0PV2rdoaR2KdYq1Wlot1TihlIrFgWJRQVEGEZQkQAIK1kKo4oAMhlEE/L1/7HVk93CGHZKTfc6T7+e69nWe/Qxr/Z51zkVu1l5771QVkiRJ0ky30bALkCRJktYFg60kSZI6wWArSZKkTjDYSpIkqRMMtpIkSeoEg60kSZI6wWArSZKkTjDYStJ6lmRVkruS3N73eMQ6aPOAdVXjAP0dn+Rf11d/E0lyVJJvDrsOScNnsJWk4Xh+VW3Z9/jxMItJMnuY/T9YM7VuSVPDYCtJ00SSOUk+luQnSa5P8o4ks9qxxyT5epKbktyY5JQkc9uxk4FHAV9ss79vTLJfkh+Nav83s7ptxvWzSf41ya3AURP1P0DtleTVSf4ryW1J3t5qvijJrUn+Pckm7dz9kvwoyV+3e1mV5IhR4/CpJD9Pcl2SNyfZqB07KsmFSf4hyc3AZ4APAXu3e/9FO++gJJe2vn+Y5Pi+9he0eo9M8j+thjf1HZ/Vavvvdi/LkuzYju2c5KtJbk5ydZIXr9EvWdKUMthK0vTxSeBe4LeBpwDPAf6kHQvwbuARwOOBHYHjAarqJcD/cP8s8N8N2N/BwGeBucApk/Q/iOcCuwNPA94ILAGOaLU+EfjDvnMfDmwLzAeOBJYkeVw79gFgDvBbwLOAlwIv67t2L+AaYHvgj4FXAhe1e5/bzrmjXTcXOAh4VZIXjKr3mcDjgN8B3prk8W3//2u1HghsDRwN3JnkIcBXgX9rff8hcGKSJww+RJKmksFWkobj9CS/aI/Tk+wAPA94fVXdUVU3AP8AHA5QVT+oqq9W1d1V9XPg/fRC39q4qKpOr6pf0wtw4/Y/oPdU1a1VdSVwBXB2VV1TVauBL9MLy/3e0u7nPOBM4MVthvgw4K+q6raqWgW8D3hJ33U/rqoPVNW9VXXXWIVU1blVtbKqfl1VK4BP88DxeltV3VVVlwOXA09u+/8EeHNVXV09l1fVTcDvAauq6hOt7+XA54BD12CMJE0h1yZJ0nC8oKq+NvIkyZ7AxsBPkozs3gj4YTu+PfBPwD7AVu3YLWtZww/7tneaqP8B/axv+64xnj+87/ktVXVH3/Pr6M1Gbwts0p73H5s/Tt1jSrIXsJjeTPEmwKbAaaNO+2nf9p3Alm17R+C/x2h2J2CvkeUOzWzg5MnqkbR+OGMrSdPDD4G7gW2ram57bF1VIy9zvxso4ElVtTW9l+DTd32Nau8OYIuRJ20mdLtR5/RfM1n/69pD20v7Ix4F/Bi4EbiHXojsP3b9OHWP9Rx6ywXOAHasqjn01uFmjPPG8kPgMePsP69vfOa25Q+vGrBdSVPMYCtJ00BV/QQ4G3hfkq2TbNTefDXy8vlWwO3AL5LMB94wqomf0VuTOuL7wGbtTVQbA2+mN2v5YPufCm9LskmSfei9zH9aVd0H/DvwziRbJdmJ3prXiT5a7GfAI0fenNZsBdxcVb9ss+F/tAZ1fRR4e5KF6XlSkocB/wk8NslLkmzcHnv0rc2VNGQGW0maPl5K72Xz79JbZvBZYF479jbgqcBqeutRPz/q2ncDb25rdo9t61pfTS+kXU9vBvdHTGyi/te1n7Y+fkzvjWuvrKqr2rHX0qv3GuCb9GZfPz5BW18HrgR+muTGtu/VwN8muQ14K72wPKj3t/PPBm4FPgZsXlW30XtD3eGt7p8C72GC/2GQtH6laqxXcCRJmhpJ9gP+taoeOeRSJHWMM7aSJEnqBIOtJEmSOsGlCJIkSeoEZ2wlSZLUCQZbSZIkdYLfPCa23XbbWrBgwbDLkCRJmtSyZcturKrRXzgDGGwFLFiwgKVLlw67DEmSpEkluW68Yy5FkCRJUicYbCVJktQJBltJkiR1gsFWkiRJnWCwlSRJUicYbCVJktQJBltJkiR1gsFWkiRJnWCwlSRJUicYbCVJktQJBltJkiR1gsFWkiRJnTB72AVo+FZev5oFx5057DIkSdIMtmrxQcMuwRlbSZIkdYPBVpIkSZ1gsJUkSVInGGwlSZLUCQZbSZIkdYLBVpIkSZ1gsJUkSVInGGwlSZLUCQZbSZIkdYLBVpIkSZ1gsJUkSVInGGwlSZLUCQZbSZIkdYLBVpIkSZ1gsB0lydwkrx52HWsjyfFJjh12HZIkSevTjAu26ZnKuucC0yLYrod7lSRJ6owZEZqSLEjyvSQnAsuBtyS5JMmKJG/rO++lbd/lSU5u+7ZL8rl2/iVJntH2H5/k40nOTXJNkte1ZhYDj0lyWZL3jlPPlknOSbI8ycokB7f97+mf7W19/EXbfsNYNQ9wrzsm+ZckS5NcOep+VyV5W18dO4/R3v9N8uUkmw863pIkSTPR7GEXsAYeB7wMOB04FNgTCHBGkn2Bm4A3Ac+oqhuTbNOu+0fgH6rqm0keBZwFPL4d2xnYH9gKuDrJvwDHAU+sqt0mqOWXwCFVdWuSbYGLk5wBnAqcAJzYznsx8NwkzwEWjq65qs6f6F6r6tUASd5UVTcnmQWck+RJVbWinXtjVT21BepjgT8ZaSTJa4DnAC+oqrv7O0hyDHAMwKytt5vgViVJkmaGmRRsr6uqi5P8Pb2wdmnbvyW90Phk4LNVdSNAVd3cjh8A7JJkpJ2tk2zVts9sge/uJDcAOwxYS4B3tUD9a2A+sENVXZpk+ySPALYDbqmq/2mzwWPVPF6wva6qLu57/uIWRGcD84BdgJFg+/n2cxnwB33XvAT4Eb1Qe8/oDqpqCbAEYNN5C2vA+5YkSZq2ZlKwvaP9DPDuqvpw/8EWHscKaBsBe1fVXaPOB+ifxbyPwcfjCHrBdfequifJKmCzduyz9GaUH05vBnfcmicwcq8keTS9mdg9quqWJCf19dV/D6PrvwLYDXgkcO2A/UqSJM1YM2KN7ShnAUcn2RIgyfwk2wPn0JvZfFjbP7IU4WzgNSMXJ9ltkvZvo7c0YSJzgBtaqN0f2Knv2KnA4fTC7WcnqXkQW9MLuquT7AA8b8DrLgVeQW/ZwyMGvEaSJGnGmkkztgBU1dlJHg9c1GZdbwf+uKquTPJO4Lwk99ELdkcBrwP+OckKevd7PvDKCdq/KcmFSa4AvlxVbxjjtFOALyZZClwGXNV3/ZVtqcP1VfWTiWoGbhjgfi9PcilwJXANcOFk1/Rd+832sV9nJvk/I8s0JEmSuihVLq/c0G06b2HNO/KEYZchSZJmsFWLD1ov/SRZVlWLxjo2E5ciSJIkSQ8w45YirE9JdgVOHrX77qraax20/TB664JH+52qumlt25ckSdrQGGwnUFUr6X2ywFS0fdNUtS1JkrQhcimCJEmSOsFgK0mSpE4w2EqSJKkTDLaSJEnqBIOtJEmSOsFgK0mSpE4w2EqSJKkTDLaSJEnqBL+gQew6fw5L19P3O0uSJE0VZ2wlSZLUCQZbSZIkdYLBVpIkSZ1gsJUkSVInGGwlSZLUCQZbSZIkdYLBVpIkSZ1gsJUkSVIn+AUNYuX1q1lw3JnDLkPT1Cq/vEOSNEM4YytJkqROMNhKkiSpEwy2kiRJ6gSDrSRJkjrBYCtJkqROMNhKkiSpEwy2kiRJ6gSDrSRJkjrBYCtJkqROMNhKkiSpEwy2kiRJ6gSDrSRJkjrBYCtJkqROMNhKkiSpE2ZMsE3yrQdxzblJFk1FPdNZkpOSHDrsOiRJktanGRNsq+rpw65hfUsya9g1SJIkzRQzJtgmuT09701yRZKVSQ7rO/7Gtu/yJItHXbtRkk8meUeSWa2NS5KsSPKKds7JSQ7uu+aUJL8/Ti0LklyQZHl7PL3t/0ySA/vOOynJC8frc5y290vyjST/Bqxs+05PsizJlUmOGTUm72z3fHGSHcZo7+2tjhnzu5YkSXowZg+7gDX0B8BuwJOBbYFLkpzf9r0A2Kuq7kyyTd81s4FTgCuq6p0tGK6uqj2SbApcmORs4KPAnwP/kWQO8HTgyHHquAH4P1X1yyQLgU8Di4BTgcOALyXZBPgd4FXAy8fqs6quHaf9PYEn9h0/uqpuTrJ5u+fPVdVNwEOAi6vqTUn+Dvi/wDtGGmn75gAvq6rq76CNwzEAs7bebpwyJEmSZo6ZNov3TODTVXVfVf0MOA/YAzgA+ERV3QlQVTf3XfNhWqhtz58DvDTJZcC3gYcBC6vqPOC3k2wP/CHwuaq6d5w6NgY+kmQlcBqwS9v/ZeDZLbw+Dzi/qu4ar88J7vM7o0Lv65JcDlwM7Nh37a+A/2zby4AFfde8BZhbVa8YHWoBqmpJVS2qqkWztpgzQSmSJEkzw0ybsc0E+x8Q3ppvAfsneV9V/bKd+9qqOmuMc08GjgAOB46eoI4/B35Gb+Z4I+CXAG0G91zgd+nN3H66r77x+hzLHSMbSfajF9z3brPR5wKbtcP39IXW+/jfv89LgN2TbDMq6EuSJHXSTJuxPR84rK1Z3Q7YF/gOcDZwdJItAEYtRfgY8CXgtCSzgbOAVyXZuJ372CQPaeeeBLweoKqunKCOOcBPqurXwEuA/jd5nQq8DNin9cUkfU5mDnBLC7U7A08b8LqvAIuBM5NsNeA1kiRJM9ZMmrEt4AvA3sDl7fkbq+qnwFeS7AYsTfIrekH2r39zYdX727rZkRnZBcDyJAF+Tm99LlX1syTfA06fpJYTgc8leRHwDfpmWOmF7E8BZ1TVr9q+j47X5wC+ArwyyQrganrLEQZSVae1UHtGkgPbsghJkqROyhjLL6edJA8DllfVTlPczxb0PongqVW1eir7mk42nbew5h15wrDL0DS1avFBwy5BkqTfSLKsqsb8noJpvxQhySOAi4C/n+J+DgCuAj6wIYVaSZKkrpj2SxGq6sfAY9dDP18DHtW/L8nvAu8Zdeq1VXXI2vaXZFd6SyP63V1Ve61t25IkSRuiaR9sh6l9isGgn2Swpm2vpPf5u5IkSVoHpv1SBEmSJGkQBltJkiR1gsFWkiRJnWCwlSRJUicYbCVJktQJBltJkiR1gsFWkiRJnWCwlSRJUif4BQ1i1/lzWLr4oGGXIUmStFacsZUkSVInGGwlSZLUCQZbSZIkdYLBVpIkSZ1gsJUkSVInGGwlSZLUCQZbSZIkdYLBVpIkSZ3gFzSIldevZsFxZw67jA3eKr8kQ5KkteKMrSRJkjrBYCtJkqROMNhKkiSpEwy2kiRJ6gSDrSRJkjrBYCtJkqROMNhKkiSpEwy2kiRJ6gSDrSRJkjrBYCtJkqROMNhKkiSpEwy2kiRJ6gSDrSRJkjrBYCtJkqROmLbBNsncJK9+ENd9KcncKShpRkhybpJFw65DkiRpfZu2wRaYCzwg2CaZNdFFVXVgVf1ibTtPMntt25jJ/UuSJM000znYLgYek+SyJJck+UaSfwNWAiQ5PcmyJFcmOWbkoiSrkmybZEGS7yX5SDvn7CSbt3Mek+Qr7foLkuzc9p+U5P1JvgG8Z6yikuyZ5FtJLm0/H9f2fzvJE/rOOzfJ7kkekuTj7R4uTXLweDec5KgkpyX5InB2ki2TnJNkeZKVI9dOdG99bW2U5JNJ3jFOX8ckWZpk6X13rh7k9yFJkjStTedgexzw31W1G/AGYE/gTVW1Szt+dFXtDiwCXpfkYWO0sRD456p6AvAL4IVt/xLgte36Y4ET+655LHBAVf3FOHVdBexbVU8B3gq8q+0/FXgxQJJ5wCOqahnwJuDrVbUHsD/w3iQPmeC+9waOrKpnA78EDqmqp7Zr35ckk9wbwGzgFOD7VfXmsTqpqiVVtaiqFs3aYs4E5UiSJM0MM+nl7u9U1bV9z1+X5JC2vSO9oHfTqGuurarL2vYyYEGSLYGnA6fdnxHZtO+a06rqvgnqmAN8MslCoICN2/5/B74K/A29gHta2/8c4PeTHNuebwY8CvjeOO1/tapubtsB3pVkX+DXwHxgh/Hura+NDwP/XlXvnOA+JEmSOmUmBds7RjaS7AccAOxdVXcmOZdeYBzt7r7t+4DN6c1S/6LNBE/YzzjeDnyjqg5JsgA4F6Cqrk9yU5InAYcBrxgpF3hhVV09Sbtj9X8EsB2we1Xdk2QV99/nWPc24lvA/kneV1W/HLBfSZKkGW06L0W4DdhqnGNzgFtaqN0ZeNqgjVbVrcC1SV4EkJ4nr0Fdc4Dr2/ZRo46dCrwRmFNVK9u+s4DXjiwhSPKUNezrhhZq9wd2GvC6jwFfojcrPZP+50WSJOlBm7bBtqpuAi5McgXw3lGHvwLMTrKC3gzqxWvY/BHAy5NcDlwJjPuGrjH8HfDuJBcCoz+h4bPA4fSWJYx4O73lCivavbx9Dfo6BViUZGmr+apBL6yq9wPLgZOTTNvfsyRJ0rqSqhp2DRqyTectrHlHnjDsMjZ4qxYfNOwSJEma9pIsq6oxP7PfmTxJkiR1gusvx5HkZcCfjdp9YVX96Tpo+3d54OfkXltVh4x1viRJkiZnsB1HVX0C+MQUtX0WvTeVSZIkaR1xKYIkSZI6wWArSZKkTjDYSpIkqRMMtpIkSeoEg60kSZI6wWArSZKkTjDYSpIkqRMMtpIkSeoEv6BB7Dp/DksXHzTsMiRJktaKM7aSJEnqBIOtJEmSOsFgK0mSpE4w2EqSJKkTDLaSJEnqBIOtJEmSOsFgK0mSpE7wc2zFyutXs+C4M4ddxgOs8rN1JUnSGnDGVpIkSZ1gsJUkSVInGGwlSZLUCQZbSZIkdYLBVpIkSZ1gsJUkSVInGGwlSZLUCQZbSZIkdYLBVpIkSZ1gsJUkSVInGGwlSZLUCQZbSZIkdYLBVpIkSZ1gsJUkSVInrHGwTXJ8kmMfxHW3r+k166KeJOcmWTQVfU+1JK9M8tJJzjkqyQfHOfbXU1OZJEnS9DMtZ2zTM/TaksweZv9V9aGq+tRaNGGwlSRJG4yBwmOSNyW5OsnXgMe1fY9J8pUky5JckGTntn+HJF9Icnl7PH1UW1smOSfJ8iQrkxzc9i9I8r0kJwLLgR2TvCHJJUlWJHnbRPVM4kVJvpPk+0n2aW1sluQTrYZLk+zf9h+V5LQkXwTOTjIvyflJLktyRd/1z0lyUbuP05JsOc7Y7Znk82374CR3Jdmk9X/NJGP5m9noJHu0cbgoyXuTXNHXzSPa9f+V5O/a+YuBzVvdpwwwRpIkSTPapDOSSXYHDgee0s5fDiwDlgCvrKr/SrIXcCLwbOCfgPOq6pAks4DRge+XwCFVdWuSbYGLk5zRjj0OeFlVvTrJc4CFwJ5AgDOS7AvcMU49E95nVe2Z5EDgb4ADgD8FqKpdW5A8O8lj2/l7A0+qqpuT/AVwVlW9s93PFq3uNwMHVNUdSf4S+H/A347R9/JWK8A+wBXAHq32b7f9441lv08Ax1TVt1po7bdb6+Nu4OokH6iq45K8pqp2G2tAkhwDHAMwa+vtxhs3SZKkGWOQl9r3Ab5QVXcCtBC6GfB04LQkI+dt2n4+G3gpQFXdB6we1V6Ad7WQ+mtgPrBDO3ZdVV3ctp/THpe251vSC7pbjVHPZD7ffi4DFrTtZwIfaHVeleQ6YCTYfrWqbm7blwAfT7IxcHpVXZbkWcAuwIXt/jcBLhqr46q6N8kPkjyeXkh/P7AvMAu4oM30jjeWtHucC2xVVd9qu/4N+L2+U86pqtXt3O8COwE/nGhAqmoJvUDNpvMW1kTnSpIkzQSDriEdHXw2An4x3mzgJI4AtgN2r6p7kqyiF5ShNxs7IsC7q+rD/Rcnef0Y9Uzm7vbzPu6/54xz7v+qo6rObyH8IODkJO8FbqEXfv9wwP4vAJ4H3AN8DTiJXrA9lsHGcqJa4f77g/99j5IkSRuMQdbYng8ckmTzJFsBzwfuBK5N8iL4zZu9ntzOPwd4Vds/K8nWo9qbA9zQQu3+9GYXx3IWcPTI2tUk85NsP049D8b59EI2bQnCo4CrR5+UZKdW70eAjwFPBS4GnpHkt9s5W/QtYxivr9cDF1XVz4GHATsDV1bVrYw/lgBU1S3AbUme1nYdPuA93tNmmiVJkjpv0mBbVcuBzwCXAZ+jN/sIvVD48iSXA1cCB7f9fwbsn2QlvZf+nzCqyVOARUmWtjauGqffs+m95H5Ra+uz9F6OH6+eNXUiMKu1/RngqKq6e4zz9gMuS3Ip8ELgH1s4PQr4dJIV9ILuzhP09W16yy3Ob89XACuqamTmebyx7PdyYEmSi+jN4I5e4jGWJcAK3zwmSZI2BLk/W2k6S7JlVd3eto8D5lXVn62Ltjedt7DmHXnCumhqnVq1+KBhlyBJkqaZJMuqaszvKHAt5sxxUJK/ovc7u47ejLEkSZKazgTbJP8MPGPU7n+sqk+sxxq+ADx61O6/rKqz1rbtqvoMvSUTkiRJGkNngm1V/ek0qOGQYdcgSZK0oRr619ZKkiRJ64LBVpIkSZ1gsJUkSVInGGwlSZLUCQZbSZIkdYLBVpIkSZ1gsJUkSVInGGwlSZLUCZ35ggY9eLvOn8PSxQcNuwxJkqS14oytJEmSOsFgK0mSpE4w2EqSJKkTDLaSJEnqBIOtJEmSOsFgK0mSpE4w2EqSJKkTDLaSJEnqBL+gQay8fjULjjtzvfS1yi+CkCRJU8QZW0mSJHWCwVaSJEmdYLCVJElSJxhsJUmS1AkGW0mSJHWCwVaSJEmdYLCVJElSJxhsJUmS1AkGW0mSJHWCwVaSJEmdYLCVJElSJxhsJUmS1AkGW0mSJHWCwVaSJEmdsMEH2yT7JXn6WrZx+7qqZ1S7X0oyd5Jzzk2yaIz9uyU5cCrqkiRJmo42+GAL7AesVbCdKlV1YFX94kFevhtgsJUkSRuMzgbbJKcnWZbkyiTHtH3PTbI8yeVJzkmyAHgl8OdJLkuyT5KTkhza187t7eeW7ZrlSVYmOXjAOk5M8vtt+wtJPt62X57kHW37j5N8p9Xw4SSz2v5VSbZt229JclWSryb5dJJj+7p5Ubv+++0eNgH+FjistXnYGHUdk2RpkqX33bl6TYdXkiRp2pk97AKm0NFVdXOSzYFLkvwH8BFg36q6Nsk27fiHgNur6u+hFzjHae+XwCFVdWsLmxcnOaOqapI6zgf2Ac4A5gPz2v5nAqcmeTxwGPCMqronyYnAEcCnRhpoSw1eCDyF3u9sObCsr4/ZVbVnW3rwN1V1QJK3Aouq6jVjFVVVS4AlAJvOWzjZPUiSJE17XQ62r0tySNveETgGOL+qrgWoqpvXsL0A70qyL/BreiF1B+Cnk1x3AfD6JLsA3wUemmQesDfwOuBIYHd64Rtgc+CGUW08E/iPqroLIMkXRx3/fPu5DFiwhvclSZLUCZ0Mtkn2Aw4A9q6qO5OcC1wOPG6Ay++lLdFIL2lu0vYfAWwH7N5mVlcBm03WWFVdn+ShwHPpzd5uA7yY3izxba2PT1bVX010S5N0c3f7eR8d/Z1KkiRNpqtrbOcAt7RQuzPwNGBT4FlJHg2QZJt27m3AVn3XrqI3gwpwMLBxX5s3tFC7P7DTGtRzEfB6esH2AuDY9hPgHODQJNuP1JVkdNvfBJ6fZLMkWwIHDdDn6PuSJEnqtK4G268As5OsAN4OXAz8nN5yhM8nuRz4TDv3i8AhI28eo7cO91lJvgPsBdzRzjsFWJRkKb3Z26vWoJ4L6K2D/QG99bHbtH1U1XeBNwNnt3q/yv3rcGnnXEJvje7l9JYdLAUme8fXN4BdxnvzmCRJUtdk8vc+aTpIsmVV3Z5kC3ozv8dU1fJ10fam8xbWvCNPWBdNTWrV4kEmmyVJksaWZFlVPeAz/MH1mDPJkvYGtM3orcldJ6FWkiSpKwy260iSXYGTR+2+u6r2WhftV9UfrYt2JEmSuspgu45U1Up63/YlSZKkIejqm8ckSZK0gTHYSpIkqRMMtpIkSeoEg60kSZI6wWArSZKkTjDYSpIkqRMMtpIkSeoEg60kSZI6wS9oELvOn8PSxQcNuwxJkqS14oytJEmSOsFgK0mSpE4w2EqSJKkTDLaSJEnqBIOtJEmSOsFgK0mSpE4w2EqSJKkT/BxbsfL61Sw47swpa3+Vn5ErSZLWA2dsJUmS1AkGW0mSJHWCwVaSJEmdYLCVJElSJxhsJUmS1AkGW0mSJHWCwVaSJEmdYLCVJElSJxhsJUmS1AkGW0mSJHWCwVaSJEmdYLCVJElSJxhsJUmS1AkG22kqyfFJjh12HZIkSTOFwXYGSTJ72DVIkiRNVwbbaSTJm5JcneRrwOPavnOTvCvJecCfJXl+km8nuTTJ15Ls0M5bmWRuem5K8tK2/+QkBwzvriRJktYPg+00kWR34HDgKcAfAHv0HZ5bVc+qqvcB3wSeVlVPAU4F3tjOuRB4BvAE4Bpgn7b/acDFU38HkiRJw+VL29PHPsAXqupOgCRn9B37TN/2I4HPJJkHbAJc2/ZfAOwLXAf8C3BMkvnAzVV1++jOkhwDHAMwa+vt1vGtSJIkrX/O2E4vNc7+O/q2PwB8sKp2BV4BbNb2n08vHO8DnAv8HDiUXuB9YEdVS6pqUVUtmrXFnHVQuiRJ0nAZbKeP84FDkmyeZCvg+eOcNwe4vm0fObKzqn4IbAssrKpr6C1ZOJZxgq0kSVLXGGyniapaTm/JwWXA5xg/kB4PnJbkAuDGUce+DXy/bV8AzKcXcCVJkjovVeO9+q0NxabzFta8I0+YsvZXLT5oytqWJEkbliTLqmrRWMecsZUkSVInGGwlSZLUCQZbSZIkdYLBVpIkSZ1gsJUkSVInGGwlSZLUCQZbSZIkdYLBVpIkSZ1gsJUkSVInGGwlSZLUCQZbSZIkdYLBVpIkSZ0we9gFaPh2nT+HpYsPGnYZkiRJa8UZW0mSJHWCwVaSJEmdYLCVJElSJxhsJUmS1AkGW0mSJHWCwVaSJEmdYLCVJElSJxhsJUmS1Al+QYNYef1qFhx35lq3s8oveZAkSUPkjK0kSZI6wWArSZKkTjDYSpIkqRMMtpIkSeoEg60kSZI6wWArSZKkTjDYSpIkqRMMtpIkSeoEg60kSZI6wWArSZKkTjDYSpIkqRMMtpIkSeoEg60kSZI6wWArSZKkTjDYrmNJTkpyaNv+aJJdJjj3qCSPWH/VSZIkddcGEWzTs97vtar+pKq+O8EpRwEGW0mSpHWgs8E2yYIk30tyIrAc+FiSpUmuTPK2vvP2SPKtJJcn+U6SrZLMSvLeJJckWZHkFRP0kyQfTPLdJGcC2/cdOzfJotbeSUmuSLIyyZ+3Wd1FwClJLkuyeZK3tj6vSLIkSfraeU+r7/tJ9mn7ZyX5+9bmiiSvbft3T3JekmVJzkoyb0oGWZIkaRqZPewCptjjgJdV1auTbFNVNyeZBZyT5EnAVcBngMOq6pIkWwN3AS8HVlfVHkk2BS5McnZVXTtGH4e0fnYFdgC+C3x81Dm7AfOr6okASeZW1S+SvAY4tqqWtv0frKq/bdsnA78HfLG1Mbuq9kxyIPA3wAHAMcCjgadU1b1JtkmyMfAB4OCq+nmSw4B3Akf3F5TkmHY9s7bebk3HVZIkadrperC9rqoubtsvbmFuNjAP2AUo4CdVdQlAVd0KkOQ5wJNG1soCc4CFwFjBdl/g01V1H/DjJF8f45xrgN9K8gHgTODscerdP8kbgS2AbYAruT/Yfr79XAYsaNsHAB+qqntb/TcneSLwROCrbcJ3FvCT0R1V1RJgCcCm8xbWOPVIkiTNGF0PtncAJHk0cCywR1XdkuQkYDMg9MLtaAFeW1VnDdjPhMGw9flk4HeBPwVezANnUDcDTgQWVdUPkxzfahxxd/t5H/f/3saqP8CVVbX3gLVLkiR1QmfX2I6yNb2QuzrJDsDz2v6rgEck2QOgra+dDZwFvKq9rE+SxyZ5yDhtnw8c3ta7zgP2H31Ckm2Bjarqc8BbgKe2Q7cBW7XtkRB7Y5ItgUOZ3NnAK1vNJNkGuBrYLsnebd/GSZ4wQFuSJEkzWtdnbAGoqsuTXErvpf1rgAvb/l+1NagfSLI5vfW1BwAfpfdy//L2Bq6fAy8Yp/kvAM8GVgLfB84b45z5wCf6Ppnhr9rPk4APJbkL2Bv4SGtnFXDJALf2UeCxwIok9wAfqaoPtiUU/5RkDr3f8Qnt3iVJkjorVS6v3NBtOm9hzTvyhLVuZ9Xig9a+GEmSpAkkWVZVi8Y6tqEsRZAkSVLHbRBLEdaFJLsCJ4/afXdV7TWMeiRJkvS/GWwHVFUr6X0erSRJkqYhlyJIkiSpEwy2kiRJ6gSDrSRJkjrBYCtJkqROMNhKkiSpEwy2kiRJ6gSDrSRJkjrBYCtJkqRO8AsaxK7z57B08UHDLkOSJGmtOGMrSZKkTjDYSpIkqRMMtpIkSeoEg60kSZI6wWArSZKkTjDYSpIkqRMMtpIkSeoEg60kSZI6wWArSZKkTjDYSpIkqRMMtpIkSeoEg60kSZI6wWArSZKkTjDYSpIkqRNSVcOuQUOW5Dbg6mHX0VHbAjcOu4iOcmynjmM7tRzfqePYTp3pNLY7VdV2Yx2Yvb4r0bR0dVUtGnYRXZRkqWM7NRzbqePYTi3Hd+o4tlNnpoytSxEkSZLUCQZbSZIkdYLBVgBLhl1Ahzm2U8exnTqO7dRyfKeOYzt1ZsTY+uYxSZIkdYIztpIkSeoEg+0GJMlzk1yd5AdJjhvjeJL8Uzu+IslTh1HnTDTA2O6c5KIkdyc5dhg1zlQDjO0R7e91RZJvJXnyMOqciQYY24PbuF6WZGmSZw6jzplosrHtO2+PJPclOXR91jeTDfB3u1+S1e3v9rIkbx1GnTPRIH+3bXwvS3JlkvPWd42TqiofG8ADmAX8N/BbwCbA5cAuo845EPgyEOBpwLeHXfdMeAw4ttsDewDvBI4dds0z5THg2D4deGjbfp5/t+t0bLfk/iVrTwKuGnbdM+ExyNj2nfd14EvAocOueyY8Bvy73Q/4z2HXOtMeA47tXOC7wKPa8+2HXffohzO2G449gR9U1TVV9SvgVODgUeccDHyqei4G5iaZt74LnYEmHduquqGqLgHuGUaBM9ggY/utqrqlPb0YeOR6rnGmGmRsb6/2rxfwEMA3ZQxmkP/eArwW+Bxww/osboYbdGy15gYZ2z8CPl9V/wO9f9vWc42TMthuOOYDP+x7/qO2b03P0QM5blNnTcf25fReddDkBhrbJIckuQo4Ezh6PdU20006tknmA4cAH1qPdXXBoP9N2DvJ5Um+nOQJ66e0GW+QsX0s8NAk5yZZluSl6626AfnNYxuOjLFv9OzLIOfogRy3qTPw2CbZn16wdR3oYAYa26r6AvCFJPsCbwcOmOrCOmCQsT0B+Muqui8Z63SNY5CxXU7vK1dvT3IgcDqwcKoL64BBxnY2sDvwO8DmwEVJLq6q7091cYMy2G44fgTs2Pf8kcCPH8Q5eiDHbeoMNLZJngR8FHheVd20nmqb6dbo77aqzk/ymCTbVtV0+b746WqQsV0EnNpC7bbAgUnurarT10uFM9ekY1tVt/ZtfynJif7dDmTQnHBjVd0B3JHkfODJwLQJti5F2HBcAixM8ugkmwCHA2eMOucM4KXt0xGeBqyuqp+s70JnoEHGVg/OpGOb5FHA54GXTKdZgxlgkLH97bTk1T4lZRPA/3GY3KRjW1WPrqoFVbUA+CzwakPtQAb5u31439/tnvSyjn+3kxvk37L/APZJMjvJFsBewPfWc50TcsZ2A1FV9yZ5DXAWvXc+fryqrkzyynb8Q/TemXsg8APgTuBlw6p3JhlkbJM8HFgKbA38Osnr6b3b9Nbx2tXAf7dvBR4GnNj+Lbu3qhYNq+aZYsCxfSG9/9m9B7gLOKzvzWQax4BjqwdhwLE9FHhVknvp/d0e7t/t5AYZ26r6XpKvACuAXwMfraorhlf1A/nNY5IkSeoElyJIkiSpEwy2kiRJ6gSDrSRJkjrBYCtJkqROMNhKkiSpEwy2kiRJ6gSDrSRJkjrBYCtJkqRO+P9o9uEUaxgZpAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# find feature importance\n",
    "feature_importance(smote_rfc, X_train, y_train)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Plot SHAP values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Exact explainer: 5287it [00:44, 97.87it/s]                                                                             \n"
     ]
    }
   ],
   "source": [
    "# SHAP values\n",
    "import shap\n",
    "\n",
    "# fit the model\n",
    "lgbm.fit(X_train, y_train)\n",
    "\n",
    "# Create object that can calculate shap values\n",
    "explainer = shap.Explainer(lgbm.predict, X_test)\n",
    "\n",
    "# Calculate Shap values\n",
    "shap_values = explainer(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x309.6 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot the SHAP values\n",
    "shap.plots.beeswarm(shap_values)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The bee swarm plot is used to understand the importance or contribution of features for the whole dataset. Looking at the Recent Average Rank variable, low values (in blue) has a very high contribution towards the prediction. All the little dots represent a single observation. The horizontal axis represents the SHAP value.\n",
    "\n",
    "All variables are shown in order of global feature importance, the first one being most important and last one being least important."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x360 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "# Plot the SHAP values\n",
    "shap.plots.bar(shap_values)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In the bar plot, the features are ordered from the highest to lowest effect on prediction. It takes in account the absolute SHAP value, so it does not matter if the feature affects the prediction in a positive or negative way."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAqIAAAFZCAYAAABDvE20AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjQuMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/MnkTPAAAACXBIWXMAAAsTAAALEwEAmpwYAAByOUlEQVR4nO3dd3wUZf4H8M+zKbvZbHpCSAIJkAAhoffeO0FEf9hQz/POB89y1jv1zn62O89Tz3IyiooUe6V3pAoCAtICpFJMQhrpZbPz+2M3sAnpZDOb5PN+vfbFzDPPPPOdLLv55nlmnhGqqoKIiIiIqKXptA6AiIiIiNonJqJEREREpAkmokRERESkCSaiRERERKQJJqJEREREpAkmokRERESkCSaiRERERKQJJqJEREREpAkmokRERETNSAgRJITYIITIEUIsEkK8LIR4sIH77hVCxDo4RKch+GQlIiIiouYjhPgPAIOqqvcIIYIAHAQQpapqcQP2vQHAjaqqXt8csSiK4g9gEYCpADIBPCGlXF5DPT2AdwFMBuAP4DSAv0kp19RQtzuAXwF8JaW81Va2FcBwAGZbtXNSyp71xcceUSIiIqLmNRnAl7blOwCsbkgSavMDgAlCiJBmiuUdAGUAggHMB/A/RVFq6nF1BXAGwDgAPgCeAvCFoihdamnz5xrK75NSmmyvepPQyoMSERER0VUSQrgDuADAG8AKIUSSbf1Duzr/AtBdVdW5tvVXAQwEMF1V1XJVVUuEEPth7cFcfDXxKIriCeB6AL2llAUAdiiK8gOA2wA8bl9XSlkI4Fm7opWKoiQBGAQg2a7NmwDkAtgFIOpq4gOYiBIRERE1C1VVy4QQIwBsUVU1GACEEBcAxNtV+yeABCFEf1iHsqcDGK2qarldneMA+tV0DEVRVgIYXUsIO6SUcXbrPQBUSClP2pUdgrXXs06KogTb9j9qV+YN4HkAkwD8oYbdXlYU5RVYz/fvUsqt9R2HiSgRERFR8+kPa7JXyRdAfuWKqqpZQog3AHwC6xD4aFVVL1ZrIx9AjUPz1RLN+pgAVG/7IgCvunZSFMUNwDIAi6WUJ+w2/QPAIinlGUVRqu/2GIBjsF4GcBOAFYqi9JdSJtR1LCaiRERERM2nP6omojm4MvH7BcAzAOarqnqmhja8YB3+vloFsF4mYM8bdolxdYqi6AAsgTWhvM+uvD+s174OqGk/KeUeu9XFiqLcDGAmgLfqCpCJKBEREVHz6QdrIlfpMKxD3D8DgBCiD4D/wXr9550ArriDHUAvAEtralxRlDUAxtRy7O1Syhl26ycBuCqK0l1KecouvqNX7gooiiJgvcM+GMBMKaX95QLjAXQBkGrrDTUBcFEUJUZKObCG5lQAopY4L+H0TURERETNRAiRBmCKqqq/2tYfBhCtqqoUQoQB2AngzwA2AkiCdaqmrXb76wH8BqC3qqrnrzYeRVE+gzUp/COsvbWrAYyUUl6RjCqK8p6tzmTbzU3224yo2rv6KKyJ6Z8AlAMYBuBHWKdvuhGAAmCglNL++tgrMBElIiIiagZCiI4AUgCYKm8+EkIEwjqPaCyAbQAWqar6X9u2RwHMVVV1lF0b8wDcrKrqdc0Rk20e0Q8BTAGQBeBx+3lEbT2s22G9JjQZQCkuzwUKAAuklMtqaPdZAFFSylsVRQmCNcGNBlAB4ASAp6SUG+qLj4koERERkQMJIV4CkKGq6hsNqLsHwB9UVT3i8MCcABNRIiIiItIEn6xERERERJpgIkpEREREmmAiSkRERESaYCJKRERERJpgIkpEREREmmAiSkRERNRChBBL6q/VfnD6JiIiIqIWIoQoVFXVU+s4nAV7RImIiIhIE65aB0DUDnDYgYiIAADz5s0D2v7vBdHgihyaJ3I4fsiIiAgAUFpaCr1er3UYjtbgRJRD80REREQt5OzZs1qH4FSYiBIRERG1EDc3N61DcCpMRImIiIhaiL+/v9YhOBUmokREREQtJD09XesQnAoTUSIiIqIW4ufnp3UIToWJKBEREVELKSsr0zoEp8JElIiIiKiFFBYWah2CU+E8okSOxw8ZEREB4Dyi1bFHlIiIiKiFcB7RqviITyKi1qjcDLzwJZDDYT4ip9Q5EHh0DiCqdg62g97QRmEiSkTUGiWkQX3lW4gys9aREFFNjO7AuFhgaPcqxd7e3hoF5Jw4NE9E1Eqpbi5ah0BEtSkpB/676oriCxcuaBCM82IiSkTUSgnR4PsBiKilWVTg65+A/OIqxQEBARoF5JyYiBIRtVIWznpC5NxcBLBsW5WioqIijYJxTkxEiYhaKfaIEjm5wlLg9RVVioqLi2up3D4xESUiarXYI0rk9M5mAQcSLq126tRJw2CcDxNRIqJWiiPzRK1AaTnw39WXVjmPaFVMRImIWimdjkPzRE6vwgJ8sRMoLAEAeHh4aByQc2EiSkTUSlks7BIlahV0OuCzHQAAo9GocTDOhYkoEVErxf5QolaisAT4j/WmpaysLI2DcS5MRJ1Yjx499gwePDipOdry8fEpmTZt2pfN0VZ7FRYWljZmzJh9WsdBVIl3zbdifibgm8eAguVA8kLg5jG1140NB9Y+BVz4GFC/qbrN3RX44B5rG3nLgAP/BqYPcGjo1ETJ6cCvKQgKCtI6EqfSJh7xqSiKDsAOACMAdJZS1nolsKIo0wG8BqAbgAQAD0sp19u2jQGwptouBgDHpJR9bXX+DGA+gD4Azkspo5r5dC45efLkMEe1TUStn0VVwWcrOblnbrT++9znVcvfuQsoMwPBdwL9uwCr/g4cSgaOnbmyjXIz8MUu4N21wPdPVN3m6gKcyQLGPQmkZgIzBwJfPAr0eRBI4RN8nEqpGXhrNfKens3HfNppKz2iDwGod4ZYRVG6AfgGwMsAfGz/fqsoShcAkFJul1KaKl8AvAGcA7DUrpnzAP4F4MVmPQNymBEjRuhtf6wQtSnsEW2ljHrg+uHAU8utQ7Y7TwA//AzcNq7m+ifPAx9uAo7WkKQWlVqT3JQL1mkUVu0HktKBQZGOPQdqvAoLsGwbyi4WaB2JU2n1v5wVRekB4B4Ajzag+u8A7JdSLpVSlkkplwE4YCuvyUwAHQF8VFkgpfxKSvk1rAmqQ0VFRR0cMGDApeMMHTq0X7du3VI9PT0tvr6+FYMGDTp21113hVZuHzJkSJfo6OhffX19zQaDQQ0LCyu+8cYbb67ebmRkpKlr165JPXr0yL3rrru6xMbGevXu3Xu7v79/uaenpyUqKipn3rx5cwCgd+/e0svLS3388cfDKvf39PT00uv1luuvv/4fdcU/aNCgKeHh4dkmk8liNBrVqKio7BtuuGEOAPTs2XOkq6urevfdd0+orC+EEL6+vsXTpk37GgDqiqsuMTExN7m4uKhjx479wN/fv+yXX34pAdChf//+HwcFBZUYDAbVz8+vYujQob88/vjjXnbHV8eOHft9SEhInsFgUDt37lw0b968m2o6hhDCpWfPnns7d+5cfNttt7HnmjTCm5VapR6h1qTk1G+Xyw6lALGdr77tDj7W9mtKWkl7OoHOe/je2GvViaitl+tDAH8BkNuAXfoB2F+t7ICtvCZ3A/haStnk8Q1FUW5RFCW3jtfKhrQjhHBNTEzc4ePjo86dO7dzdHR0n3PnzoX+8ssvu2zbdWfPnt0nhAiZNm3aiNjYWJdu3brNs1gsKfbt9O/fv1thYeFZPz8/XHvttZ3ef//95OLi4p/Ky8t7T506dWR0dLRRr9dv37Zt29d//etfg44ePfqBTqcrPXz48CuVbXTr1u0ffn5+6rRp016qK2YXFxcRGxv72uzZs/379u3b0WKxZO3Zs+cLRVHc4uPjdwUGBmbHx8c/W1k/Jibm5tLSUkPv3r3/BgB1xVXfz8tisSA7Oztu5syZsf379/cFcMFoNJ4dM2bMNbfffrtLr1695h47dqzPvn37vrDfLykpadqYMWPm9ujRw+jm5nb24MGDH1Zvu3fv3r7h4eHJQoio6dOn91iyZMmeumLJz8/nMpebfbmwsJDziLZWJgNwsdog3sVCwOsqp/VxdQGWPQgs3grEO7yvhJqioARln25ziu8QRy83VGu/RvQBAGlSym8qh9fr4QXgYrWyXACx1SsqitIZwAwAE68mQCnlcgDLr6YNAOjZs+fMhIQE0+9///tRr7766jkA53r37v3I4cOHP3j77bdDunfv3jchISHgxhtvHP7GG2/8bNutSpKbk5PTLSUl5Xjv3r333XbbbWOklBa9Xh9YVlYWc+edd85btGjRzwAghLhWr9eXJyQk3K+q6tOxsbErTp8+PbeynQsXLtwWHR29S0pZ53PK9u7dux7A+sr1iIiI+ampqXuSkpL6AtgfFBT06cmTJ+9WFMVNSlleUlLyWK9evc699tpr8fXFBeDp+n5mw4YNu2/RokWnKtellE/abf4hOjp6a1pa2lD7ffr06fPlF198scm2/HpiYuK7iqL4SCkvAkBBQYHfhQsXznTu3Dlt5syZPZ9//vl6Lwnx8vLiMpebfdnT0xMq5xF1Tiv+BozuZV02uFn/fTDO+u+O48BTnwLe1abw8TYC+Vfx6EchgCUPWK87ve/9prdDjmUyoPzawfB1gu8QRy83VKvtEVUUJQrAIwDua8Ru+bBeG2rPF0BeDXXvAhAvpfyxSQE2M3d39z5eXl7o3r37scqy1NTUHeXl5Th27Fhfg8Ew0NvbGzExMQdqa+PEiRP9DQaDS3R09G1SSgsAhIeHDwCATz/99Euj0agajUbVw8OjwmKx6AoLC7sDQElJyV8SExM977zzzut69eo1Kisryz86Ovrv9cU8aNCgwVFRUSk+Pj5mg8GgZmRk7AGAjIyMSNu/T2dkZLhs3rz5z56enl5nzpzp061bN6WhcdVFCIGBAwfutS8bPHjwS6Ghofmenp4WDw8PNTExcVJxcXGV3wYeHh6XZimoqKjIKSkpAax/wAAAEhISuhQXF5t69ep1Z0OSUCJH4jyiTmr2S4DfbdbXK99aX5Xrs1+yXvPpqgOiQi7v06/L1Q2nL7oXCPYBrn8VMFdc9SmQg6gqLHOHax2FU2m1iSiA0QCCABxRFCUT1iF2ADisKMo9texzCMDAamUDbOWXKIriCuAPABZebZCKosxXFKWgjlf1u/RrVFZW9mt+fj4SEhKiK8vCw8NHurm5ISoq6teSkpIDeXl5OHHiRP/a2hg2bNh3RqPx1Lp1647ecccdfQAgNTX1EAD8/ve/n1pUVCQqX2VlZWLt2rU3A0BCQkJyp06dTp88efIpi8XyUnR0dPZ77723rb6Y09PTv/P09HSbOnVq35KSEtGhQ4dhAKCqqs62Pbtz587HEhIS7u3Wrdsz/v7+6tSpU//V0LjqIoSAm5ubpXK9Y8eO3Q4cOPBE//79l1177bW+xcXFolu3bpvQyKkY+/Xr90tYWNimtWvXbrn55punNWZfoubG/tBWqqgU+GYP8PxN1huXRkYDc4YAS+ro99C7Wadqqr4MAP9bAPTqBMx+GSgpc2zs1HQuOuC28cgp5s1K9lrz0PwXADbarXcCsBvAVAAnatnnEwB/URTlZgBfAfg/AIMA3F6t3mwAfrb6VdiSVFcAbgCEoigGAJBSltR0QNsNUcsadkq1i4+PXx0QEFC4YcOGFefPnx95+vRp36ysrP/07ds39eGHHz7/yCOPpHXs2DF73bp1q2666aYZp06d+sVoNE4PDQ3N+fzzz3cDgE6nsyQmJsZGRUXtXb9+/f5bb711XGlp6e7IyMijW7ZsWT5v3ry5X3755Y7x48eHlZWVyb59+y557733TgOAyWT657Fjx5TS0lLLlClT3mxIzGaz2cPV1TUvNDQ0ediwYREuLi6fVa9jNBr/ffz48Q+9vb3/2LNnzz2VP8fS0tKMhsTVUHq93k9VVRiNxt+WL1+en5iYeNPZs2fHBwcHmxvTDgAcP358cu/evb9Zu3bt6htuuOHmL7744ov69yJqfrxrvhW7RwE+vBfI+AjIygf+pFyeuqlzIHDsTSDmAeBMJhARZJ0ntFLJ50ByBtD1biA8CLh7mjUBTVt0uc6ChcDyevsLqCW5uwL3zUBwsL/WkTiVVpuISimLYDdlky1BBKzXjBbYyuYDWGibiglSygRFUa6DdR7RDwEkApgrpUyu1vwCAJ9LKXNqOPSTAJ6xW6+8qMehvxFUVTUPGzZs3IULF77/5ptvzrm5uandunU7NWDAgCm27ZahQ4cOvnjx4qrVq1fvKSsrcwkICCgJCwu7E9YEvbIdC4DB0dHRW9avX7/95ptvnmUwGEa4uLh8s3nz5k0Gg8HdYDBYQkNDz7u5uV2aturo0aMf+vj4vOXh4aEfO3bsM1dGeKXw8PB7k5OTF7333nuFPj4+5ZGRkR8nJSXdZV/n6NGji729vd9LT0/3iYuLe85+W0PiaqiUlJT9ffr0+WHt2rVPGwyGZ7OzszMjIiJ+LikpadLMz0eOHLmub9++H61du/az66+/3ufrr7/mRVnU4jiPaCtQff7QSjkFwNx/1rztTCbgNf/yesoFQFxXc93UOraRc4nqCMSGIzs1FSaTSetonIZQedul04qMjDzs5+fns2/fvgitYwGsTxbq1q1b6vbt24fWX5vs8ENGze/EWahD/gpRUONgDBE5E5MBeOuPwB0TkZCQgMjINj/Pa4M751rzNaJtWo8ePfQFBQXhJpMppf7ajjdgwIBbMjMzg6Ojox/ROhYislL5Nw5R62BRgRtGAQA6deqkcTDOhYmoExJCBKSmphabTCZERkbeq3U8wcHB2fHx8cvGjBnz7fvvv7+9sjwmJuZ6vV6v1vQaM2ZMnXNrXi29Xl9R03F79uzZ+EnMiFopDmgRtQIuOuDm0dYb0wCcPVvrU8jbpVZ7jWhbpqpqFpzoj4T09PQar6w+duzY19Doxt3S0lJeGkftno7ziBI5P70bcP/MS6u8PrQqp0l2iIiocTiPKFErEB4I9Ot6adXVlX2A9piIEhG1Upy9icjJeeqBh6+pUpSbm6tNLE6KiSgRUSslOKU9kXOzqNbrQ+107NhRo2CcExNRIqJWysK7lYicl04H/N8IwORRpTgzM1OjgJwTE1EiolZKV85nihM5Lb0r8OdZVxRbLJYaKrdfvGKWiKg1iuyIvIemw0fl1ziRU+oUAAy6cuJ6Ds1XxScrETkeP2TkEO3kCS1EbUo7+dzyyUpERG0dn9BC1Pp4e3trHYJTYSJKRNRK8QktRNTaMRElImqlPD09tQ6BiBopLy9P6xCcChNRIqJWyt3dXesQiKiRQkNDtQ7BqfB2SyIie5l5QFK61lFU5W0EeoZdUZyTkwN/f38NAiKipkpLS0PXrl3rr9hOMBElIrI36wXg6BnA1UXrSC4rKQd2vAgMjqpSHBwcrFFARNRUOh0Ho+0xESUispdfAhSWah1FVTod8N9VwCcPVCnOzs6GyWTSKCgiaorAwECtQ3AqTMuJiJydxQJ8tRsoKK5SXF5erlFARNRUaWlpWofgVJiIEhG1BjoBfLqjShHnESVqfXx9fbUOwakwESUiag0KS4H//FCliPOIErU+ZrNZ6xCcChNRIqLWIjUTOJh0aZXXhxK1PgUFBVqH4FSYiBIRtRal5cBbqy+turryflOi1oaX1FTFRJSIqLWosACf7QCKrHf15+bmahsPETUaL6mpiokoEVFrohPA59abljp27KhxMETUWG5ublqH4FSYiBIRtSYFJcDrKwAAmZmZGgdDRI3Fp6FVxUSUWtStt976sU6ns7TkMW+77baFer2+wRMuzpkzZ4e3t/dFR8ZEbZCfCfjmMaBgOZC8ELh5TN31H4wDflsE5C4BFt0LuNuu93R3BT64x9pG3jLgwL+B6QOq7ns6DTiaCoulRT9KRNQM0tOd7BHCGuOV7o2kKMpNAO4F0A+AUUrpWm17PwCvAOgPoCOAMVLKHdXq3A7gMQDhALIBfATgOSml6vATaIQhQ4YUpKWl/XzmzJkJWsdyNZYsWbJgyZIlC5qrPUVRxi9YsGALgDGqqu6odwdqW5650frvc59XLX/nLqDMDATfCfTvAqz6O3AoGTh25so2pvYHHr8OmPgMcD4b+PYx4LmbgCeWWh8teiYLGPek9S75mQOBLx4F+jwIpFyw7l9mBt5ajY6v/85x50lEDuHn56d1CE6FPaKNlwPgXQAP1rK9DMA3AK6paaMtUf0QwBMAvAFMA7AAwB+bO1AiaiFGPXD9cOCp5UBhCbDzBPDDz8Bt42qu/7vxwKJN1iQ1txD4x5fAHba/94pKrUluygVAVYFV+4GkdGBQ5OX9KyzA0h/xW1KKw0+NiJpXWVmZ1iE4FSaijSSlXCel/BRAYi3bj0sp35dS/lxLE5EAMqSUP0gpVSnlCQBbYe1hbXazZ8/+NiQkpNRgMKje3t7myMjIXUIIz8rtf/7zn2OGDRsWHxAQYDYYDGpQUFBReHj4zClTphw7cOCA5/nz58fr9XrV3d3dLIRwmThxYnJwcPBp+2P07NlTjYiIUADg6aefHtarV68L3t7eFoPBoAYHB1+MiYm5vbFxjxo1KjM8PHwjACiK4hEVFWXx8vK6NPna1KlTT4SEhBy3bXedM2fOkrCwsBKj0agGBgYW9erV6++VdatfDvDMM890GjZsWKKXl5fFy8vL3L9//y87dOighoWFPVpZRwghpkyZssvX19diNBotXbt23SaEcFEUJfT555/fDABubm7b9Xq9Gh4e/mNjz4/amB6h1uTw1G+Xyw6lALGda64fG27tLb1UNxno6Af41zAvaAcfa/tHq/WsCoHAraeuNnIiamGFhYVah+BUODTf8tYBOK8oylwA3wOIATAWwO9r20FRlJUARtfR5j1SyuU1bTCbzafCwsKmhYaGbouNjb1lw4YNn4SHhy8CcJOiKMZt27btTUtLywsJCRk4Z86cY6dPn5574MCBcxs2bIipaWh+4sSJdZ5ceXk59Hq90qNHj/+MGzcOhw4d2rx9+/YPhRBrVFW9UOfOdsLCwnb++uuvEwAgLS1t6pkzZ2A0Gg1CiB4LFy5MSk1NjXJzc3sDALZu3frJhg0b5kVFRc2fMGHCN6qqPvvtt9/+Iyws7Jdz586trt72jh07fjx16lRQp06d+k+aNCl33759u7KysqrcgVxQUOCVkJBw1t/f3zh58uQ7ly1b9k5YWNi9Usr/Api4YMGCLeXl5RyaJyuTAbhYVLXsYiHg5dGw+pXLXh5Att1k164uwLIHgcVbgfhzVdsoKIHHoi3APbOvNnoiakGcR7Qq9oi2MCllIazXhC6GdRj/VwBLpJTr69gnTkrpW8erxiQUANasWfPXffv2bd2/f7/lk08+WTpgwIBDhYWF4wAgMTHxlkOHDnnm5+dPP3LkyOEXX3zR/Pnnn3956tSpQ009v5dffnnPwYMH/75v376s1157LeuGG26Y7+bm5uLt7V1XIn2FmJiYJYWFhV5CiG5Hjx693dfXN7lnz54FnTt3vj03N3dcYmKiy7lz5xYriiIOHz58Q1BQ0CeHDx/+YunSpeZly5Y9OWDAgGydTveX6u0qitLp119/7ebh4fGP48ePH3777bdT58yZc5dOV/Wj4Orqak5KSro5MTGxRFGUd7t3716m0+kmNeVnkp+fz+VWtFxR0w1AK/4G5Cyxvh6fa31Vrq/4m/VOdm9j1X28jUB+8ZVtAbb6dklq5bJ9fSGAJQ9Yrwe97/0rmlCFQEmIT7OeO5e5zGXHL589e1bzGFpiuaHYI9rCFEX5PYBnAUwB8DOsNyx9pijK81LKp5v7eNdff/07+/fvvyMzM9PDbDYLVVXh7u5eAADp6en9AKCoqOh4cx3viSeeiNmwYcOKpKSkiMLCQhedToeSkhKEhIR0bUw7oaGhayMiItTy8vJbzp07N8ZgMHwdFhbWNTU1dc6hQ4ciXF1dS8rKyn5VFCUoOzvbJSsr606j0Xhn5f4WiwV+fn4hNTQdlpubCyHEL5UFAQEBp6o/KlGv1xeVlpZWVK67ubmVq6rq3ZhzqOTl5cXlVrTsoqvh7/PZL11erulmJaMecNUBUSHAadvwfL8uVw6nVzqaat3+5a7LddNyqvaGLroXCPYBZr4ImCuuaEIY3eH2l7lXfb5c5jKXW3ZZr9drHkNLLDcUE9GWNwjAZinlHtt6sqIoywDcCaDGRFRRlDUAxtTR5gIp5bLqha+++uq4lStX3hMeHv5aaGjoMydPniycOXPmL9u3b+8OAMHBwYcAwGAw9ARwpPr+Qogruobc3NyKLBaLj11sobm5uTAYDACAn3/++YekpCS/rl27Dtm/f//BhQsXxj788MO/opG971LKgk8//fRsYmLiTYmJif5ZWVkfRUdH91y1atWHZ86c8ff396/stc308/OrcHNzeyclJeWBBjR9ztfXF25ubgMAbACArKysqIKCAhiNxnp2vcQihICqOtUkB6SlolLgmz3A8zcBf3wX6N8VmDMEGPm3mut/8iPw8X3Asm3AbznAk/OAj7dc3v6/BUCvTsDkZ4GSWm5sCPLBb51NaNRfeESkOW/vJvVptFlMRBtJURQXAG4A3G3rBtumUimlqiiKAKC328XdVqdcSlkBYCeANxVFGSSl3K8oSmcAtwI4UNsxpZQzmhJrSUmJf3l5Ocxm87knnnii+NZbb737wIED/QAUAUC3bt0+7du375sZGRlr+/TpM2Pq1KnHzp07N3f//v2nTp06dchoNOaVlZV1EULoVFW1AEBQUNDenJyc23v06DFpzpw5+w8fPvx1ZmbmpWteysvLjTqdznz+/PmkBx98MGT58uVfFBcXw8fHp/ZAa9GpU6eNu3fv/r0QoqysrGxfRETEyaCgIN3PP/8c6u/v/4btZ6Nu2bLli5UrV949YMCAnw4ePPjFXXfd5X3hwoVbTp06lXDkyJG11X6WZz/77LPkgwcPPhkdHb16+vTpF3/66af3GzkfY5q3tzc8PT1HAOA1omR1jwJ8eC+Q8RGQlQ/8Sbk8dVPnQODYm0DMA8CZTGDdL8C/vgO2PA94uANf/wQ885m1bngQcPc0awKatuhy+wsWAsu3WZeN7sCDcdC5uLToKRLR1btw4QKTUTu8RrTxbgNQDOtNRy625WIAEbbtEXZlALDJtnwbANjuuH8V1uH4fAB7Ye2NfLi5Aw0ODv5+2rRpP+bk5Lx2//33V2zZsuWZ7t27/1q5XUpZOG7cuOGdO3cuOXfu3MG3337bvGnTpsUlJSWhANC9e/en9Xp9iIeHh1mv15cLIVzGjh37wLBhw06eP39+w7vvvpuVmJh4Lioq6tIxe/fufZePj49HVlZWzocffphy8eLFXwMCApoUf9++fT8GAH9//yOqqlqklLldu3bNKCkpwfnz5z+rrDdhwoTbJ02a9ElWVtYiDw8P87Jly7J+/vnn58rKymocIxgxYsS4yMjIrLNnzx5etGhRUllZ2T5fX19UVFQ06FZGKeXJUaNGbS4qKnrFaDRaIiIiNjfpBKl1eu7zK+cQBYCcAmDuPwHTLUDEAuDT7Ze3nckEvOZb/630+gqg452Az63AnW9brwUFgNQLgLgO8LjJuk/lqzIJBQCLCtw+HoGBgY45RyJymKb+TmyrBIcXqb17/vnn5zz33HPfARhVUVGxywGH4IesNYn5M3D8rNZR1E4I4LphwFd/RUJCAiIjI7WOiIga4fz58wgNDdU6DEcTDa3IoXlqdx544IHR27Zt65Oenv7BnDlzYvbu3fuByWTKz8vLq23uVyLn4akHHpwNAPD19dU2FiJqtOLiWmbTaKc4NE+aUBQlPCQkpEKv16vVX35+ftmOPLarq2vAhQsXXs/Ozi5bvHjxLxkZGaXBwcGTVVVt8PPoiTTj5wmMigYAmM1mjYMhosbiPKJVcWieyPH4IWtNnHlo3sMdeP5m4NE5AMCheaJWqJ18bhs8NM8eUSKi1sKiXn4mPdizQtQaeXjU8sS1doqJKBFRazGtPxB4edqXs2edtOeWiGrViDmr2wUmokRErYHJADwYV6XIzc1No2CIqKmysrK0DsGpMBElImoNvAzA+N5Vivz9/TUKhoiaKigoSOsQnAoTUSIiZ2dwA+6fZZ1D1E56erpGARFRU+Xl5WkdglPhPKJERPbGxcKSmQedaPBNn47n6wn8YdIVxX5+fhoEQ0RXo7S0VOsQnAqnbyJyPH7IWpmzZ8/yjnQicojS0lLo9Xqtw3A0Tt9ERNRU7LEgIkfhbBdVMRElIqqGvaFE5Cienp5ah+BUmIgSEVXDHgsichR3d3etQ3AqTESJiKphjwUROUpOTo7WITgVJqJERNWwx4KIHCU4OFjrEJwKp28iopbz3OfA22u0juKyV28H7ph4RXFOTg4niycih8jOzobJZNI6DKfBRJSIWs7O40CmE03m/ORy4HcTrpgonj0WROQo5eXlWofgVDg0T0Tt18UiYPuxK4qzs7M1CIaI2gPOylEVE1Eiar8KS4HXV1xRzB4LInIUzspRFRNRImq/VBVY8wuQnV+lmD0WROQovD60KiaiRNS+6QSweEuVIvZYEJGjuLry9hx7TESJqH0rLgPeWGntHbVhjwUROUpubq7WITgVJqJERFn5wO74S6vssSAiR+nYsaPWITgVJqJEREVlwJsrL62yx4KIHCUzM1PrEJwKE1EiIlUFfvgZyC0EwB4LInIci8WidQhOhYkoETk3PxPwzWNAwXIgeSFw85i66z8YB/y2CMhdAiy6F3C3DbO7uwIf3GNtI28ZcODfwPQBl/fT6YBlPwJgjwUROQ7/0K2KiSgROYdnbrS+qnvnLqDMDATfCcx/HfifBGI619zG1P7A49cBk54FutwNdAsGnrvJus3VBTiTBYx7EvC5FXjqU+CLR4GIIOv2olLgPysAVWWPBRE5zPnz57UOwakwESVqoJiYmO+6d+9epHUc7YpRD1w/HHhqOVBYAuw8YR1Cv21czfV/Nx5YtAk4dsY6zP6PL4E7Jli3FZVan3WfcsE6FL9qP5CUDgyKvLx/ei6w7zR7LIjIYby9vbUOwam0iVtDFUXRAdgBYASAzlLKWicBVBTlbgAPAwgFcArAQ1LKrTXU6wtgH4BtUsrJduV/BjAfQB8A56WUUc14Kk7Px8enZPjw4SvWrVs3T+tYqB3oEQpUWIBTv10uO5QCjIupuX5sOPD9z3Z1k4GOfoC/CcguqFq3g4+1/aNnLpcVlwFvrsL552YjMjISRETkWG2lR/QhAPX2VCmKMg/APwDcAMAHwEIAqxRFCa9WzxXAhwC219DMeQD/AvDiVcbcaHFxcR4tfcy2QAgh5s2bZ9A6DmoCk8H6PHh7FwsBr1o+CtXrVy5Xr+/qAix7EFi8FYg/d7ncogLf/AQfnfvVRk5EVKO8vDytQ3AqrT4RVRSlB4B7ADzagOrzACyVUh6UUlZIKd8DkAHgjmr1ngDwM2pIRKWUX0kpvwZwrvq25hYWFpY2ePDgE926dUsxGAxqXl7eFgAYOHDgOx06dCj28PBQQ0NDS2bNmvWK/X5Dhgx5MiQkJN/Dw8Pi7e1tGT58+KHKbcOGDbu9U6dOOUaj0RIQEFA+ZsyYTc8++6wbAMTExNzk4uKijh8//m1/f/8yDw8PNSYmJu0Pf/hDFABEREQk5efn6zdv3vx/er1e7dGjR2Fd8QshjJGRkcleXl4VBoNBDQ0NLZ41a9ZLtm2uXl5eZXFxce/b7xMVFXVwwIABly6gqe9c6zi2Onr06NUhISF5bm5uFgCPDhky5K6wsLA8o9FoMZlMlpiYmPPz588fYv/zHjJkyLHIyMgkg8GgBgQEmGfMmPFubccYMGDAa76+vhVz5859qK5Y8vPzuVxt+ZIVfwNyllhfj8+1virXV/wNKCgBvI1V9/E2AvnFV7YF2OrbJZ2Vy/b1hQCWPGC97vS+91GdarFA2BJYZ/hZcZnLXG5by6GhoZrH0BLLDdWqh+ZtQ/IfAvgLgNwG7KIDIKqVCQD97drsA2tiOgDWIfyrjfEWALUmMwB2SCnjatt49OjRnpMmTfrH3Llznztx4oRvnz59Xk9LS7t75MiRf7BYLEvS0tLu37hx4+u33nrroaVLl37av3//P504ceIf48ePfyMkJOTv8fHxJpPJdCMA9O3bd8zJkycXT5gw4X+dO3d++PDhw0OPHDmywWg0fgbgesA6rUROTs41cXFx3ePj411PnTp15OTJk58CGJKSktK1kUPzuk6dOm0YOnToC8nJyWkFBQWLNm/e/MQf//jHVaqq7uzVq9fWxMTEGwDcBQAmk8lUXl7eNy4u7kUAqO9c6zt4QkLC5FGjRl1/+PDh1UIIk6ur67UDBgz4a1BQ0CfHjx8PO3PmzK7Dhw+vARBYuc/x48d7TZw48fGYmJh/Jycnv//jjz/+6ZFHHnnztddei7dvOzY2duWFCxemTZs2be7nn3/+Q11xeHl5cbna8iWzX7q8XHmj0nOfXy4z6gFXHRAVApy2Dc/361J1ON3e0VTr9i93Xa6bllN1WH7RvUCwDzDzRcBccUUTol8X5Hq5IqAZz5fLXOYylyuX09LS0LVrV6eJx1HLDdWqE1EADwBIk1J+oyhKlwbUXwHgNUVRlgI4BGABgHAAp4FLQ/IfAXhQSpmnKMpVByilXA5geVP3j46O/m3lypVP21azAgMD/zBo0KDvvv/++49tZW9EREQ8kJKS8giAT7Ozsx/v37//ybVr11b20hUBeAsASktLX+7Vq9dva9asuce2bVvv3r0/tSWDl4wePfoP77zzTort+DuysrKGoAlUVS2ALcm0udVoNN6YkZFxC4Cdqqo+derUqZ8WLFgwceHChZu7du36dHp6ujpt2rSXAeC3336r81zrO37fvn1XfPXVVytsqxcBLLbbfKpPnz7PpKSkvKMoiqeUstB2vik//PDDPwFACHE/gN+fP39+OoB4AKioqBBRUVFHKyoqusyYMWPQ4sWLDzflZ0MNVFQKfLMHeP4m4I/vAv27AnOGACP/VnP9T34EPr4PWLYN+C0HeHIe8LHdc+T/twDo1QmY/CxQUnbl/iYD8MgchIaGOuR0iIh0ulY/GN2sWm0iqihKFIBHAAxuxG6fAOgIYBmsvWDfA9gEIMu2/a8ATkkpV9S8e8vz9PRMs1/Py8sz/fjjj9cZjcZLD8auqKhAYGCgDgAKCwv9TSbTtpraKi4u7pSUlBRiv6/FYoGfn9+luWp0Oh369et33G69sLy8vEkXzHXq1Mno7e29NS0trX9RUZGbTqdDSUkJSkpKQgDgxIkTezp27JgVHx//LIDNWVlZv4uOjt4npSxqyLnWx9PT84T9+rBhw+akp6crWVlZgeXl5ToAKC0tBaz/FwoBwMPD49IEkqqqFgohUFZWFlBZlpGRoS8uLo6ZPXv2v5iEtpB7FODDe4GMj6yP4vyTYr0rHgA6BwLH3gRiHgDOZALrfgH+9R2w5XnAwx34+ifgmc+sdcODgLunWRPQtEWX21+wEFhu+8gIAHOGIu382So9FkREzSUwMLD+Su1Iq01EAYwGEATgiK3nsjI5OawoypNSyiuGw6WUKoB/2l5QFMUdQCKAF2xVpgIYqChKZTJiBOBqW+8hpcxubJCKosyH9aao2myXUs6obaMQosqEhl5eXoXDhg37bvXq1bfVVN/T0zO7oKAgsqZtBoPhfN++fbFv374uDQi9tnjU+mtZ+fj4fJyTkzNgwoQJcQEBARsURbEYjcZy2F2bHBQUtOzkyZP3Dhw4cEh6enqHmTNn/rFyW33n2oBYq4y7njx58tOePXsmDB06dOgXX3yR0qdPn3uOHDnyDq68XKNWISEhJd7e3i+sX7/+hdmzZ6srVqx4vCmxUQ3sh+Tt5RQAc/9Z87YzmYDX/Kplr6+wvqpLvQCI62o/vqsLcMdEQO/GHgsicpi0tDTOymGnNSeiXwDYaLfeCcBuWJPJEzXtoCiKD6zTNp2AtRfsFQB5uDxkOw+A3m6XhwEMAXAzbNeg2obvXQG4ARCKohgAQEpZUtMxpZTLYO2BbRZhYWEfHDhw4L5rrrlmc/fu3T/Zs2ePpxBiXmBgYO633377tZ+f36uHDh16a8aMGa8GBAQ8nZiY6GUymW5Yv37920aj8Ynjx49vmTFjxptBQUF/P3r0aKmnp+cEDw+PfuvWrXu1Icf38PAoKCgo6N6QuhaLxcfFxcXi6el5+tChQ4bevXsvKy0trfJ/7ty5c88UFBTc7+HhsSoqKqrggw8+uJRB1HeujfvJAeXl5W7u7u4FnTt3PjtixIh+2dnZzze2DQDYv3//S4MHD764adOmt2bMmOG7Zs2au5vSDjkZNxfgXuvfhOyxICJH8fX11ToEp9JqE1Hb8O2leVpsCSJgvWa0wFY2H8BCKaXJts0bwJcAugAoA7AKwAQpZbGtzQv2x1AUJQ9AabV5SZ8E8IzdeuXtuA3uVbsahw8ffmjQoEHq3r17396wYcMiFxcXBAcH53bo0OERADh06NDbgwcPDvrll18eyc3NfdTd3d0SGxt7GMDbBw8e/HH48OHzjx079t/MzMz7AcDPz68kNja2wYlyWFjYC8ePH/+X0Wi0dO7cuTA+Pr7WK5NdXFwW6HS6PZ9//vlpDw8PS58+fbZ7e3tXud05Ozs7NzIy8lhiYmLsjBkzPmnMuTZWdHT0k0ePHn1+z5495sDAwJKQkJBvz58/f3NT2tq3b987Q4YMubht27bFU6dO9V2/fv1NTWmHnEh0GNAzDAB7LIjIccxms9YhOBWhqg0eaSWipuGHrNLUZ4ENTnhprclgvZHpVusTm7KyshAQEFDPTkREjZeQkNAe/tBtcOccL4QiIlJV4P9GXFpljwUROUqnTp20DsGpMBGlq6LX6yv0er1a/dWzZ8/Gz2rbCOHh4ak1HVev16vVn5RFVCcXHXDbeMBweXKIgoKC2usTEV2Fs2drfQp5u9RqrxEl51BaWuqixXFTU1OZbFLzcHcF7qs6cQV7LIjIUdzc3LQOwamwR5SI2reojkBs1b9r2GNBRI7i7++vdQhOhYkoEbVfJgPw0OwritljQUSOkp6ernUIToWJKBG1XxYVuHH0FcXssSAiR/Hz89M6BKfCa0SJqOV4GaF6uEO4O8FXT5kZuGUMYNRfsSk9PR0mk6mGnYiIrk5ZWZnWITgVJ/htQETtxkf3IWPHQQQHB2sdiVWfmu95Y48FETlKYWGh1iE4FU5oT+R4/JDZaSeTORMR1ai0tBR6/ZUjMW0MJ7QnIufEqZGIqD3jrBxVMRElohbFL2Eias/aQW9oozARJaIW5enpqXUIRESa8fb21joEp8JElIhalLu7e/2ViIjaqAsXLmgdglPhXfNEdHU+2gwcSbmy/PoRwMjoK4pzcnI4TycRtVsBAQFah+BUeNc8keO17Q+Z1y1AQcmV5V07AAn/A0TVmycLCgo4RycRtVvnz59HaGio1mE4Gu+aJyKNZVwEfj59RXF2drYGwRAROYfi4mKtQ3AqTESJyDGKy4A3V15RXF5erkEwRETOgVPYVcVElIgcw6IC3+wB8oqqFPNLmIjaM05hVxUTUSJyHBcBLNtWpYhfwkTUnnl4eGgdglNhIkpEjlNYCrxRdXieNyoRUXtmNBq1DsGpMBElIsc6mwUcSLi06urKWeOIqP3KysrSOgSnwkSUiByrtBz47+pLq7m5udrFQkSksaCgIK1DcCpMRInIsSoswBc7gQLrlCUdO3bUOCAiIu3k5eVpHYJTYSJKRI6n0wGf7wQAZGZmahwMEZF2SktLtQ7BqTARJSLHKywB/rMCAGCxWJreTnY+MPcVwPNmIEICy7fVXf/1FUDHOwGfW4E737ZeJlDp7dXA4L8A+huAO95qekxERI3AKeyqYiJKmnJxcVEnTpz4IACMHj36nY4dO/KRE21Vcjrwa0rDhuaf/cz6qu7e9wF3VyD9Q2DZQ8CfFOBoas1trPsFeOUbYNOzQPJ7QGI68Ixdm6H+wJP/B9w5qUmnQ0TUFJzCrirevlqNoig3AbgXQD8ARinlFT8jRVEiAfwbwERb0XEAY6SU5YqidAGQBKAIl58xniul7GS3fwcA7wGYAqAEwIcAnpBSXkVXUeu3Y8eOe2H92dfL39/fHBMTo+zYseMeB4dFzaXUDLy1Gucfm4rIyMjG719YAnz9E3DkDcDkAYzuBVwzBFjyI/DKbVfWX7wV+MMkIDbcuv7UPGD+G5frXjfc+u++BOud/URELcDT01PrEJwKE9Er5QB4F4AHAKX6RkVRggBst227A0ABgAEAKqpV7SmlrO3PnmUA8gF0AhAAYC2AbAD/vPrw6xYXF+excuVK9jpSy6uwAMu2wfvJa5q2/8nzgIsO6BF6uaxfBPDjsZrrH00F5gyxq9sFSM8FsvKBAK+mxUBEdJXc3d21DsGpcGi+GinlOinlpwASa6nyMIBUKeWzUsqLUsoKKeW+hvZmKorSFcBkAH+x7Z8IawJ6d7OcQDURERG5/fr1+6V79+6/GQwGNSMj45sZM2YM7tat2wWTyWTx8PBQw8PDL06ZMmW+/X5jx479Z2hoaKGHh4fq7e1dMXDgwB2V2yZNmnRveHj4RU9PT0tAQED5sGHDfpg9e7aoL5ZZs2aFxMTEnPb09LT4+fmZR48eXSXRHzVq1AcBAQFlletjxox5s0OHDqUGg0H18vKqiImJOQ0AUVFRabm5uS579uz5k16vV7t27ZpZWb9jx47FBoNB9fb2rujdu/exWbNmXZonw9/f3zxkyJB1Xbp0ydbr9WpwcHDJpEmTZOX22bNni5EjRy4JDg4uMRgMqq+vr3nEiBGfV24fN27cy6GhoYVGo1Ht0KFD6ejRo99p3LtB0Am47U9q2r4FJYBPtYmgfTyB/Fr+rqpev3K5tvpERC0gJydH6xCcChPRxpsA4JSiKN8ripKtKMphRVHm11Bvj6IoFxRF2aooyni78n4ALkopE+zKDgDooiiKd00HVBTlcUVRcut4vVtXwPHx8f07d+786pQpU3RBQUG3qqrq2rlz54Xjxo0LmjRpkr+fn1/ivn37FsfFxXkAwPjx45/au3fvX3v06PHvSZMmGcaMGRPu5+e3DACmTp16zc6dO9/u2rXr2xMnTtQPHjz4mlOnTs3Iycl5r74fXHJy8tb8/PwOY8eO7Tdy5MiuGRkZs2q7cWXWrFkBu3bt+nPv3r2fKCkpEWPHju0YEhLyOgCcPn26o6+vb8WwYcP+V1paKpKSkgIBwMXFJbNv3743TpkyxXXEiBFTMjIyIn/77bfv7dtNTEyc0KNHjzumTp3q3rFjx2OHDh16u3JbZmbmsqNHj97Su3fvh6ZMmeI6cuTIGE9PzzUAMGHChMf27dv3WK9evf46adIk1759+y44cODAnyZOnHhffeedn5/f5pcbzKIiu0dgze3EvQjV91bA91bglW+hvvKtddn3VpinPweYDEBeUZV9Sy7kAF4eNcdmMgB5xZfL82wJaA31y8vLq+7LZS5zmcsOWg4ODtY8hpZYbiihqmr9tdohW/K4sfo1ooqinAbQFcCNAL6DNTFdAWCylHKHoigmAL1hTS7dANwJ4F8AhkkpDyuKchuAF6SUEXZtdoW1B7ZzHcP5TRIREZFrMpkyjh492qO2OrNmzQpZvXr1+SlTplyzfv36FZGRkRleXl7JBw8eHFq9bt++fQ9XVFQY7NsbNmzYDxkZGaOSkpICajvG7NmzXdauXWsePXr0X7Zs2fJvAJg2bdrk9evXb5gwYcJDmzdvfmPUqFEfnDhx4vasrCz3WbNmBWzYsCFz0KBBn/v5+f1t9erVVXqoG3KN6IgRI75ITU2dee7cOVPlPlFRUWv37t0bB1iT6g0bNnw/c+bM8FWrVp3p0KFDaWRk5He7d+++sXpbUVFR6b6+vkf27ds3ye5ncaiiokJ/9OjR6NpisGnbHzKvW6y9j/Vx0QF3TkLxm3fU/6zlyhuVnr3pcllhCeB3O3D0DaC7bXj+9jetNx3VdI3oLa8DXTsAL9r+Ttz8q7Us7cOq9Z5cbr1G9OP76z8HIqKrlJqaivDwcK3DcLR6R0kr8RrRxssHsFtK+ZVtfYOiKGsBXANgh5SyAMBPtm1lAN5SFOUaAPMAHLbt71OtTV+7tpudh4fHGfv1mTNn9khOTl7z22+/RRQXF7vodNaOcbPZ3BUA8vPzfYKDg0/U1FZhYWHHM2fOBBmNxkvJlcVigY+PT50To5nN5miz2Qx3d/f9lWVubm7ba6u/atWqrIkTJz6Umpr6+KFDh06HhoYWR0ZGfrR9+/ZaeyAnTJjwWEJCwt+ys7O9zGazUFUVnp6eZvs67u7ul26x1ul02bb4gwGcyc3NdTcYDAdqajs/P9/nzJkzE6ufd1hY2IW6zpvs6F2B+2ciLS0NXbt2bfz+ngbgumHA058BH9wDHEwCvv8Z2PVSzfVvHwfc8TYwfywQ4ge88CVwx4TL280V1leFxfoqKQNcXawvIiIHsR+BISaiTXEQQFQN5XX1ellw+a+DQwB8FEXpZrs+FLDe7JQspbxY086KovwNwN/qaH+plLLWa0yFEFXGv1NTU38oLi72HT58+KA1a9YcquwRhe1SDS8vr4vFxcU9a2rLaDRmREdHpx8+fLhPHfFcwdXV9YSrqyvKysoGAdgCAOXl5WPq2mfz5s1vAHhj9uzZbnl5ef/Yvn37Y9OmTft63bp1W1Dt5x0XF+e5a9euVwYOHPhF37597165cmXOiBEjvoiPj7+uoTH6+vqWlZSUDKhpm8lkyouIiNhb2ZtKTdAlGOgTAV1KStPbeFcCd74DdPi99Yaj/8nLd8WnXgBiHgCOvQmEBwHTBwJ/vRaY8DRQXAZcPxx4zq6H9YUvgee+uLy+9EfgmRuq9sISETUzziNaFRPRahRFcYF1SN3dtm6wbSqVUqoAFgLYrijKtQB+ADAOwFTY7nhXFGU4rHfSn4D153u7rc7fAEBKmaQoykYA/1IU5U5Y75p/zNZujaSULwGopdun8crLy42urq5lLi4uybNmzQpOTU3dYL+9c+fO7/7000/PjB8//imTyfQvVVUDS0pKZm/atOm9kJCQJ3fu3PnNuHHjnvfy8noZgKWsrGyq2Wzuunnz5v/WdswVK1ZUxMbGnk5ISHhqxowZa3Q6XW5SUtLHtdWfMWNGbGFh4R1eXl7/XbVq1ZmJEyemA4AQohwAPD09i4uLi2Mq66uqaiovL4erq+uFlStX5kydOnX26dOn5zTm5xIZGfnd8ePH/2/SpEl3GY3GDysqKiLLyspGbNy4cXF4ePh/Dhw48OKkSZPuNRqNiqqq7qWlpdeqqqrbuHHjksYcp13yNAAPzwYABAYG1l+/tmTQ3wv47vGat4UHAQXLq5Y9fI31VdsxmHQSUQs7e/Zs06awa6N4s9KVbgNQDGAdABfbcjGACACQUv4E4BZYE898AG8B+J2Ucrdt/66wXjt6EcA5W3uzpZT7Lx8C82H92Z8D8DOA72G9jrRFdOnSZUFxcbFpw4YNubt37z7r4+OzrXJ4HgC2bt367NChQ9+Ij49/YuPGjSXbt29Pzc3NvQUA1q9f/93w4cMXpKSk3Ld169bCLVu2FB89evRTs9lc7wUvXbp0Ge/p6Zn1448//rpr166UDh06rLY/bjWuqampf9iyZUuKwWBQjxw58srIkSMXrl27dgcAREZG/js5OXmU0Wi0dOvW7cKqVavShw8fvvTXX3+9W6/Xq/Hx8Z9ERETsqK3xmgQGBt7Uq1evr3/99de3N2zYYN69e/exwsLCGQCwZcuWVwYMGPCP+Pj4VzZt2lS6ZcuW/Pj4+HcsFkut18WSHYsFuGk0ACAtLU3jYIiItGMymbQOwanwZiUix2vbH7L6blZy0QG3jQM+st4MlJWVhYAA5u9E1D61k+/ABt+sxB5RInIsvRtw/8xLq2azuY7KRERtW25urtYhOBUmotRshgwZsk6v16s1vSZNmtSgR3dSG9QpABh4+XqogoICDYMhItJWx44dtQ7BqfBmJWo2P//88zStYyAn46kHHqw60QDvGCWi9iwzM5PPm7fDHlEicpwK1TqPp52zZ5v1mQ1ERK1KbU8UbK+YiBKRY+iEdQJ676rPh3dzc9MoICIi7XFoviomokTkGB7uwANXzv/v7++vQTBERM7h/PnzWofgVJiIEpFjdPABhlz5ELL09HQNgiEicg7e3t5ah+BUeLMSEV2d//4ROFLDYzuvHwGIK6eS8/Pza4GgiIioNeCE9kSOxw8ZEREBABISEtrDIz4bPKE9E1Eix+OHjIiIAADFxcXw8PDQOgxH45OViIiIiJxNWlqa1iE4FSaiRERERC1Ep2PqZY8/DSIiIqIWEhgYqHUIToV3zRORQxzLVFFsvrzerwPgqmvwZUNERG1SWlpae7hZqcGYiBJRs1ufbME131pgcLGul1QA/xqnw58HMhElovbN19dX6xCcChNRImp2r+2zoLQCKK2wrusEUFiubUxERM7AbDbXX6kd4TWiRNSs0gpV/HhG6yiIiJxTQUGB1iE4FSaiRNSsFh22NHwCOSKidqZTp05ah+BUmIgSUbOxqCre+kVFSYXWkRAROaezZ89qHYJTYSJKRM1mS6rKa0GJiOrg5uamdQhOhYkoETWbN/arKGAiSkRUK39/f61DcCpMRImoWVwoUrEhRdU6DCIip5aenq51CE6FiSgRNYuPj1jA+eqJiOrm5+endQhOhYkoEV01VVXx5oGqT1IiIqIrlZWVaR2CU2EiSm2WTqez3HrrrR9rHUdrk12sYu53FfB8w4yIhWYsP26pd5/tZ4GLpS0QHBFRK1dYWKh1CE6FT1Yih1IUZTKAFwD0BlAC4Asp5T3aRkV1uXeTBe4uQPo9LjiYAcz6pgL9ggRiA2sfd39jv4V3yxMRNQDnEa2KPaLkMIqijAfwFYB/AwgA0AnAB1rGVCkuLs5D6xicUWGZiq9PqvjHKB1M7gKjOwlcEyWw5FjtvaLZxSpWJ6ngbUpERPXjPKJVsUeUHOllAO9JKb+yKzvgiAN5enp6BQcHb01LSxtgMBjUESNGbLDfHhER8bHBYLi9oqJibXp6+tTw8HAAcO3ateue7OzsgaWlpa4+Pj7mIUOGrL7mmmuuk1JWhISEHO3du7fbhg0begCAyWQqDAkJ8fjLX/7iKaUsDg8PX9+pU6fBu3btajNzcZzMAVx0QA//y72f/YIEfjxTe5r5yVELXHiTEhFRg+j1eq1DcCpMRMkhFEXxBDAUwDpFUQ4ACAdwBMCjUsp9tewzGsDKOppNlVL2rWmDv7//ClVV+86fP3/6zp07Dxw+fPhHVVWrpEenT58WEyZM6DhjxoyOiYmJKgAEBATsGzt27N27d+8+7OLi8pe1a9e+HBYW9hSAZ93c3FampqY+qiiKfsGCBV28vb0NZ8+eVdPS0qYA+OHixYvDhg0btr3RPxwnVlAO+LhXLfPRA/lltSei7xxUUcSblIiIGsTb21vrEJwKh+bJUfxg/f91F4A7AIQCWA9gtaIovjXtIKXcIaX0reNVYxIqhND99ttvY0aOHPnZ+++/v/7YsWOZGRkZc6vX8/f3xw033HDtW2+9lblq1aosANi3b9+9ixcv/uXkyZMVx48ff8XHxyf97NmzswHg3LlznyQmJuqysrImdO7c+baePXsW+vj4JBw9evR3QojwgoIC7169ei2t7weRn5/vNMvjPzND/Lvm14ilpTC5AXllVfe9kFcKL3dRa5udveo+/0qlpZfvZtL658BlLnOZy1otX7hwQfMYWmK5oYSq8souan6KovgAyAXwopTySVuZAJANYL6UcnVzHUsIEQwg7b777nvirbfeesVW5gqgfP78+YuXLl16R0RExMceHh6/e/jhh3VSStVWR9ejR48vs7KyZhYWFuqFEKK0tBTjxo1L3rx5c1cA8PT0LJozZ873W7dujR02bFjGgQMHjnbu3PmW5OTk19zc3F564oknfKSU9X3yWs2HrLBMhd/bFTj6exd097Mmn7evrkCoCXhlrEuN+6xIsGD+Sgvy67hZSSeAF0br8MQw/u1LRO1bbm4ufH19tQ7D0Rp8wRaH5skhpJQXFUVJRs1JWI2JmaIoYwCsqaPZFCllbA3lF1xcXNTs7OxL21xcXLpWVFRUqSSEUCuTUAAIDQ29Nz09/bq4uLg/BwcHK6+99lppQEBAGuw+QH5+fgdTU1PH5+TkBPTo0eONlStXHiwrK7tXr9ffEhERca4BSWir4ukucF13gad3WvDBVB0OXgC+P61i1y01J6EAMKOrgCvzSyKiBikqKmoPiWiDMRElR3oXwAOKonwK4CSAh2GdwmlXTZWllNsBmBp7EFVVLZ07d96xe/fuG6WUi3/88ccDISEh39R3Z6JOpwvU6XTw8fFJ/OqrrypWrVr1Qm5ubjCAlMo6Fovly/379/8nICDAEhkZ+Z3ZbM7Jzc2tANBn9OjRnzQ21tbg3ck63LnOgg7vViDAA/jfFF2dUze56gQW9BP4z34VZRW1ViMiIgDFxcVah+BU2I9BjvRvAB8C2AwgE8AMADOklBeb+0BZWVlxAI4tXbp0/YULFy707dv3nBCiziHxs2fP/tPLy+vookWLVuTk5JT5+vreGBwcXCV7/e233z4vKSlBt27dMqSU2aqqqv7+/kcqKirQp0+fxc19Hs7A30Pgu2tdUPigK1IXuOKWXvV/TdzdT8cvEyKiBuA8olXxGlEix2sXH7LRn5qx81zN23iNKBGRVUJCAiIjI7UOw9EafI0ofysQUbN4eJAOXu711yMias88PPg8FXtMRImoWcyOFPxCISKqh9Fo1DoEp8LfG0TULNxcBP7QR8CN3ypERLXKysrSOgSnwl8ZRNRs7hmggwu/VYiIahUUFKR1CE6FvzKIqNlE+gr0DdQ6CiIi55WXl6d1CE6FiSgRNauHB/OmJSKi2tg/7piYiBJRM7s2SoCzwhER1YzziFbFRJSImpXeVeCO3gKuDZ5Fjoio/ajvqX/tDRNRImp2Dw7SIdQLCPSwvvwNQD9en09EBE9PT61DcCp8shKR4/FDRkREAIDs7Gz4+/trHYaj8clKRERERM4mJydH6xCcChNRIiIiohYSHBysdQhOhYkoERERUQvJzs7WOgSnwkSUiIiIqIWUl5drHYJT4c1KRI7HDxkREQGwTmiv1+u1DsPReLMSETm/ed9X4Ewe83Qiaj84j2hVTESJSBNHM1V8dUrFmXytIyEiajkmk0nrEJwKE1Ei0sRbv1gAAK78FiKidsTV1VXrEJwKfwUQUYsrMatYekyFO7+BiKidyc3N1ToEp8JfA0TU4r46aYEA4MGOASJqZzp27Kh1CE6FiSgRtbjX96ko4AwmRNQOZWZmah2CU2EiSkQt6mS2imOcz5mI2imLxaJ1CE6FiSgRtah3D1pQwe9hImqnODRfFRNRImoxZRUqPjyiopyJKBG1U+fPn9c6BKfCRJSIWswPp1XwYW5E1J55e3trHYJTYSJKRE2SXaxi7ncV8HzDjIiFZiw/Xn8353/2WXiTEhERXcJElIia5N5NFri7AOn3uGDZLBf8aYMFRzNr7+5Mvqjil4wWDJCIyAnl5eVpHYJTYSLahgkhwvV6vSqlnK11LG1BTEzMd927dy/SOg5nUFim4uuTKv4xSgeTu8DoTgLXRAksOVZ7r+h7hyzgpaFE1N6FhoZqHYJTaZXTSSuK0hHAmwAmwnoOvwB4SEp5qI59pgN4DUA3AAkAHpZSrrfbrgIoBqr8rgyTUl60bf8QwBQAPgAKAawB8IiUMqcZTw0AEBUVddDLyyvol19+CbuadlRVTQUgmiksoktO5gAuOqCH/+X/Xv2CBH48U3OPqNmiYuEhFWUVLRUhEZFzSktLQ9euXbUOw2m01h7RdwH4A+gJIBjAPgArFUWpMelSFKUbgG8AvAxrIvkygG8VRelSrepUKaXJ7nXRbtt/AERLKb0B9AJgBPBOM55To8TFxXlodexKPXr0cJFSumkdR32EEGLevHkGreNoSwrKAR/3qmU+eiC/rOZEdHWiigrepEREBJ2utaZejtFafxpRAL6UUmZLKcsALALQCUBALfV/B2C/lHKplLJMSrkMwAFbeYNIKY9IKQvtiiywJsLNqnfv3kuTkpL6HT58OFSv16t6vV597bXXomJiYr6Liooqio2N/dHLy6vi1KlTGQDQo0ePQ76+vmaDwaAGBweXTZky5ZPKtoQQXYQQ6l133TUXsA4tR0VFFfXu3Xu9yWSq8PLysowYMWKfoigulfsMHz58cpcuXdI8PT0tfn5+FUOHDv3lz3/+s599e+PGjfs0KCioOCkpyazT6abUdi5CCGNkZGSyl5dXhcFgUENDQ4tnzZr1km2bq5eXV1lcXNz79vtERUUdHDBgwKW5LQYOHPhOhw4dij08PNTQ0NCSWbNmvdKQn6MQQh09evTqkJCQPDc3NwuAR4cMGXJXWFhYntFotJhMJktMTMz5+fPnD6ncJywsLG3IkCHHIiMjkwwGgxoQEGCeMWPGu7UdY8CAAa/5+vpWzJ0796G6YsnPz291y2OWlUL821zja/SnZpjcgDy7pDM/Px95pYCXu6ixzTP5Ksy1ZKKFRYVX1Ocyl7nM5ba6HBgYqHkMLbHcUEJthXOpKIpyG4BbAcwHUADgRQBDpZRjaqn/HYBkKeWDdmVvAugspbzOtq4CSAPgBuvQ/T+llN9Ua+dxAH8HYIJ1GP/W6nXs6t4Ca89tbXZIKeNq2lDT0HxMTMx38fHxc0aPHr2tU6dOs3Nzc91WrVqVNXz48PfDw8PfOHjw4AkvL69nDx069OQNN9wgly9f/r4QoguApD/+8Y/Xvf/++99WtjF+/Pg14eHh//frr7/O+OWXX76aO3fu81999dUzPXv27HzmzJnk4cOHrwkJCbklISGhU3Jy8q6oqKjEHTt2DKxsr1u3bgVDhgyZfOrUqUPDhw8X77zzTnFN5yGEMI0dO/b10NDQF5KTk9MKCgoWnT59ev78+fNHf/DBBzt79eq1Xggx7NixYz4AYDKZTOXl5XlxcXEvfv3110/16dPn9bS0tD+PHDnyDxaLZUlaWtr9hw8ffn3evHm3LF269NM6frYQQqgdO3YsHzVq1PWHDx9e3b9/f9OZM2euDQgI8AgKCvrk+PHjYWfOnNnl7+8vfv3110DAmohevHgxeOLEiY+rqvrv5OTk9xMSEn7/pz/9Kfq1116Lj4mJ+a68vHzqqVOnjLGxsSsvXLgwbcKECdd//vnnP9QVC4DW9yGrR2GZCr+3K3D09y7o7mcdiLh9dQVCTcArY12uqJ9VrKLTexUoqTY07+MOrJ/ngqEhvIKEiNqHhIQEREZGah2GozX4S71VXiMKYCesvZkXAFQAOANgRh31vQBcrFaWCyDWbn2yrV0AmANgmaIoc6WUaysrSClfAfCKoihdAdwJ4HRtB5RSLgewvCEn01B+fn7q/Pnzp0opSyvLfvrpp7vsqjwVHBx8b3p6+jwA71/ZAtChQwfLjTfeOFtKWQHg6+Dg4JysrKyxAODm5vZCaGho6ebNmysT5GN9+vR58eTJk/+y7zXt16/ffz/77LM99cWrqmoBAPv4bjUajTdmZGTcAmCnqqpPnTp16qcFCxZMXLhw4eauXbs+nZ6erk6bNu1lAPjtt9/+MGjQoO++//77j237vxEREfFASkrKIwDqTEQBoG/fviu++uqrFbbViwAW220+1adPn2dSUlLeURTFs7K3Ozo6OuWHH374JwAIIe4H8Pvz589PBxAPABUVFSIqKupoRUVFlxkzZgxavHjx4friaIs83QWu6y7w9E4LPpiqw8ELwPenVey65cokFAACPASmdhH4IaHN5eRERI3i6+urdQhOpdUNzSuKogOwEcBJWK/3NMLaI7pdUZTgWnbLt9W15wvg0hwKUspNUsoS2+tzAEth7XG9gpQyCcAKAKtt8bQIX1/fUvskVAih69Onz9rAwMBSDw8P1Wg0qpmZmX6lpaVBtbVhMplKbUkoAMDV1bXMbDZ7AUBZWVlUSkqKh9FoVCtfp0+f/tfFixcB4NIzyfz8/H5pSLydOnUyxsTE7PX39y8zGAyq0WhUS0pKXEtKSkIA4MSJE3sCAwOz4uPjnwWArKys30VHR++TUhYBQF5enunHH3+8zj6etLS0LkVFRbWenz1PT88T9uvDhg2b06VLl/TKSwVOnTr1jm0YIbCyjoeHR2blsqqqhbafS0BlWUZGhj4pKSmmX79+b7fXJLTSu5N1KDYDHd6twM0rK/C/KTrEBtb+R/CDgwRMTn9FMRGRY5nNZq1DcCqtsUfUH0BXAG9JKSsTyQ8URfkngOEAvq9hn0MAJlQrGwBgUx3HsaDurmVXAGEAPGFNdKtQFGU+gIV17L9dSllbL26Ns9wIIap0J/Xp0+eZ1NTUaePHj7/NZDJ9tnTpUnNwcHC2qqpNGud0d3dPioyM7HvixAmvmrYvWLCgCwDodLoG3fvs4+PzcU5OzoAJEybEBQQEbFAUxWI0Gsth9wdQUFDQspMnT947cODAIenp6R1mzpz5x8ptXl5ehcOGDftu9erVtzXlfIQQVeI8efLkpz179kwYOnTo0C+++CKlT58+9xw5cuQdNGIIISQkpMTb2/uF9evXvzB79mx1xYoVjzcltrbA30Pgu2tr7gGtyfjOAiZ3cEJ7ImrXCgoKEBxcW79Z+9PqElEpZaaiKCcB3GO7ZrMUwO2wDr//WstunwD4i6IoNwP4CsD/ARhk2w+KovSGtWf1IKzX880CcBuAm2zbOwCYDuAHKWWuoig9APwL1us8a7wy13ZD1LKmnKObm1taXl5e9COPPKJ/7bXXSmurp6pqgE6nUz09PU8fP35cDBw48I3MzEy/qKio1KYc12KxPHX27NmbJk2a9HVYWNiCo0ePZptMpmHu7u4zNmzY8HQT2vNxcXGxeHp6nj506JChd+/ey0pLS6v8nzt37twzBQUF93t4eKyKiooq+OCDDyqH0hEWFvbBgQMH7rvmmms2d+/e/ZM9e/Z4CiHmBQYG5n777bdfNzae8vJyN3d394LOnTufHTFiRL/s7OznG9sGAOzfv/+lwYMHX9y0adNbM2bM8F2zZs3dTWmnvRFC4P4BAv/YrV5xrSgRUXvRqVMnrUNwKq1uaN7mWlh7RVMAZAG4F8A8KWUiYO2NVBSloLKylDIBwHUAnoR1OP5JAHOllMm2KkEAPgKQAyDDtv1OKWXlTSgqgDsAJCqKUghgA4AjsCa0zc7d3f1vxcXF5f/73/+KjUaj+tprr9V4VXNmZuZjfn5+SV9//fXukydPlup0uskhISHpTT3usWPHknr37j01KSlp6Lfffpt+7NixipMnT24pLy8f1JT2XFxcFuh0utzPP//8dHx8fL6fn5+ft7d3lRubsrOzczt37nwsMTExKDIyssqNX4cPH34oLCzsrb1797797rvvlh88eDD3/Pnzr+p0uiY9qDc6OvrJo0ePDnz77bfNqampP4WEhKyvf6+a7du3753Y2Njbt23bdtfUqVM/a2o77c0f+uja3p1bRESNcPbsWa1DcCqt8q55olaGHzI7074yY32ydZl3zRNRe5Oamorw8HCtw3C0Bn+pt9YeUSJqpR4apIMXb1oionbK399f6xCcChNRuip6vb6icuJ9+1fPnj0bP6ttI4SHh6fWdFy9Xq8qitLm/9RszaZECOhb3dXpRETNIz29yVfQtUn8dUBXpbS0tOG3TTej1NRUJputlItO4E/9BP71M69YIKL2x8/PT+sQnAp7RImoxcl+1q8ePn+eiNqbsrIyrUNwKkxEiajFdfISGB7COUWJqP0pLCzUOgSnwkSUiDTx0GB+/RBR+8N5RKvibwIi0sSsbtbZPXScuYmI2hHOI1oVb1YiIk246gQu3u8Cbz0zUSJqP/R6vdYhOBX2iBKRZpiEElF74+3dpIcDtllMRImIiIhayIULF7QOwakwESUiIiJqIQEBAVqH4FSYiBIRERG1kKKiIq1DcCpMRImIiIhaSHFxsdYhOBUmokTkFPactyCtkI9aIqK2jfOIVsVElIicwvDlFjy5w6J1GEREDsV5RKtiIkpEmrOo1p7QLt6czomI2jYPDw+tQ3AqTESJSHPHs6z/Bhm1jYOIyNGMRn7R2WMiSkSa++k3a49oYWGhxpEQETlWVlaW1iE4FSaiRKS5zSnWRFSnc9E4EiIixwoKCtI6BKfCRJSINLfjnApXHVBSwmlNiKhty8vL0zoEp8JElIg0lVeq4rdCwE0HmM1mrcMhInKo0tJSrUNwKkxEiUhTe9NUeLhal339/LQNhojIwTiPaFVMRIlIU7vOqSgqty7n5uRoGwwRkYNxHtGqmIgSkaY2pKgw2x6o5O6u1zYYIiIH8/T01DoEp8JEFICLi4s6ceLEB6+2nenTp48WQqgzZswY3AxhNVhsbOyJmJiY0/XVi4mJOR0bG3uiJWJyJt26dbswdOjQVQ2t7+/vbx49evS7joyprcouVjH3uwp4vmFGxEIzlh+v+0lJqqril4zL666uvGueiNo2d3d3rUNwKq71VVAU5SYA9wLoB8AopXSttr0fgFcA9AfQEcAYKeUOu+0eAD6xbY8E8LSU8oVqbQwD8CqAvgBKAawH8KCUkpNt0VVLTExs1rkyBg8evDUrK6tPUlJSQHO22xbcu8kCdxcg/R4XHMwAZn1TgX5BArGBNT8xKSEXUO0eL19UVATAq0ViJSLSQk5ODvz9/bUOw2k0pEc0B8C7AB6sZXsZgG8AXFPLdhXALgASwN7qGxVFcQGwEsBOAEEAegEIBfDfBsTWpsTFxbWq5361tnjJsQrLVHx9UsU/RulgchcY3UngmiiBJcdq7xXdfV6Fzi5H9fb2boFIiYi0ExwcrHUITqXeRFRKuU5K+SmAxFq2H5dSvi+l/LmW7SVSytellFsAlNRQxQdAIICPpJTlUspsAF/A2gPb7GbNmhUSExNz2tPT0+Ln52cePXq0Ur3OuHHjXg4NDS00Go1qhw4dSkePHv2O/faJEyf+OTw8/KKnp6fFZDJZYmNjT9V0rClTpsyLiIjINZlMFqPRaOnWrduFadOmTajcHhMTc7pXr15JsbGxpzw9PS0pKSk/N+T4o0aN+jAgIKDcYDCovXr1SrRYLA3u57dYLO69e/c+YTQaVR8fn4qRI0cutd8+duzYVzt27Fjs4eGhduzYsXjcuHGv2B33g4CAgLKhQ4eu9PX1Nf/88885s2fPFgMHDtzh4+NTYTAYVH9/f/Pw4cO/rNxn6tSp13br1i3TZDJZ/Pz8zAMHDtxVXwI7c+bMHjqdTp0xY8YAAJgwYcKjQgh11KhRiwAgLi5ObzAY1MmTJ//OVj8qNjY23tfX12wymSw9e/Y8M2PGjNjK9iIiInKHDBmyoXJ98uTJd4aGhhYaDAY1PDz84uDBgzf5+/tXmTeotLQ0skuXLtl6vV4NDg4umTRpkgSAMWPGvHHw4MFxqamp/nq9XtXr9er06dPHNfTn35adzAFcdEAP/8uZZb8ggaOZte+z9YyKgvLL63yyEhG1ddnZ2VqH4FTqHZp3NClltqIoCwHcpSjKkwC8AdwE4Nva9lEU5XEAj9fR7HIp5T01bUhOTt6an5/fYezYsf10Ol3uqVOnfrJYLvfYTJgw4bF9+/Y9Nnz48PuNRuN7xcXFt+3atevDiRMnHt+8efPbU6ZMuX779u1vDh06dFHfvn0fAuBaVFR0Ry1xWCIjI5/r06fPB6qq+p4+fXrX0aNHf4Dd2OOpU6e6jBgx4s1u3br9xWKxeNd3/EmTJt37888//37UqFGPe3p6vpGbm/vf3bt3y549eybU8fO4JCEhoeuIESNe7Nq16zMXL158aefOnX+dPn36u2vXrt01adKku3bv3v3oqFGjnjaZTP8qKCh4YufOnc9Mnjz55MaNGz8EgNzcXLfS0tLQ0aNHB6mqqisoKHgsMTFx+IgRI0asXbt274wZM2LNZnM/AJg5c2bMzp07v+nXr9/ymJiYBRUVFdGHDh3aptPpVgKYVFuMq1evPhkSElKcn59/N4AFFy9evC4gIKA8KytrEgAUFxf/3sXFRfXw8Fg6e/ZsceTIkf1eXl6/jRo1qiuAouTk5F3Hjx/fDuCKsY9Zs2Z13r179wexsbGrBw4cOK+0tHTO3r17l7m4uKj29RITEycMGjTo//r06bMmNTV1z6FDh94GoGzfvv3BwYMH9+fQ/JUKygGfan8S+eiB/DK15h0A/JJRdVtFRYUjQiMichrl5eX1V2pPVFVt0GvhwoXjFy5caK6njrpw4cLRdWzfunDhwidrKJ+0cOHC0wsXLjTb2ti0cOFCY0Nja+grLi7OxdXVVR0/fvyjlWVTp06dDECdMGHCg6qqIjIyMn3QoEGb7Pfr06fPoZiYmBOqqqJ3795HevTocaam9qdNmzYagDp9+vTBNW2fMmXKtQDUmTNnBqmqil69ep2OiIjItq9T3/FjY2NPREdHJ9pv79SpU16vXr1O13f+vXr1Ot2tW7cM+zJPT0/L2LFjX7S1fTw6OjrJfnvPnj1TYmNjj6mqipEjR37g5uamzpo1y6ty+4QJEx709PS0jBs37u+zZs3ysd936NChK6qf35gxY1719/cvqy/Wfv367e3Vq1eCqqoIDQ0tHD169Bsmk6kiLi5ODBo0aHP37t3PqaqKSZMm3VY9phkzZkTZvw/h4eG5gwcP3qCqKkaNGvWur6+vOS4uTlTWHzhw4DY/P79L/7f9/PzMQ4YMWWn3vl1je986q6qKQYMGbe3SpUtWQ//f5eXlqZVa8/K4T8tVvFrza/iSEvVAmkX1eL28yr4vbi9S474219rm71ebL7Xh8Xq5+u8dznO+XOYyl7nsiOWSkhLNY2iB5QbnZpr3iCqK0h3AGlivIV0GwAjrjUtrAYxtzmOZzeZos9kMd3f3/ZVlbm5u2+3r5Ofn+5w5c2ai0Wi81FVjsVgQFhZ2AQCKioo6+Pj4nGzI8aZNmzYhKSlpaXp6esfS0lKdENYhy4qKiu4ALgCAp6dnlYHLBhw/wM/P77j9PiaTqY7Bz6oMBkOVMQF3d3dLRUWFn63twOptG43G1IsXL/a0O5Z55cqV+ZXrmzdvfmPkyJGDEhMT/5qRkfFCt27dsrp16/bQxo0blxQXF3c5d+6cn/25qKoK+x7o2vj5+X134MCBf8ycObNLdna2ccCAAU8Yjca7S0pKbsjIyBgUFha2CgDKysr6m81mbN68Oc9oNF7a383NDWVlZQMB7LNvt6ysrIuXl1fhihUrLsWk1+sTAYys9nNJrVzW6XTZAGCxWIIBnKk3+Gq8vLzaxPLWm+r6unBFYZkKswVIM5sudfmfyHNHbGDtbY4PF/jy5OXh+eLiyzcraX2+XOYyl7nsiOWzZ88iMjLSaeJx1HJDaZ6IwnotaI6U8mPb+kVFUd4CcFhRFB8p5cXqOyiK8jcAf6ujzaVSyrurF7q6up5wdXVFWVnZIABbAKC8vHyMfR2TyZQXERGxd+/evXE1NWw0GjMKCwsjGnJiJ0+e/MrDwyNn1KhRPVevXn166tSp127YsOFbVL02t0pWVt/xPTw8soqKiqo8lqGgoCDQy8sro6b6jWE0GjOrt11UVNTZw8PjUqIrhLhinHXXrl23AcCsWbMCzp8/v3r//v0fAVhiMBhSw8PDgxMSEjo0NhZPT8//FRUVvZiRkfFBWFhYxsqVK4v79et3PDs7+0/nz5/37tWr14cA4O7u/qu7uzsmTZrkumLFior62nV3d0/Oz8+fOnv2bFGZjJaWlnZtZHj1Hqc98nQXuK67wNM7Lfhgqg4HLwDfn1ax65bap2QaHiJg/x9Kr+c8okTUtplMJq1DcCoNmb7JBYAbAHfbusG2qVRKqSqKIgDY//Zwt9Upl1JW2PbRAxCwJmCutu0VUspyWHusfBRFuRXAp7D2iN4HILGmJBQApJQvAXipsSe7YsWKitjY2NMJCQlPzZgxY41Op8tNSkr62L5OeHj4fw4cOPDipEmT7jUajYqqqu6lpaXXqqqq27hx45KOHTs+t3Xr1i9Gjx79nq+v7yMA3IuKin63efPmN6ofr7y8XG8ymUpcXFxSZs6c2SM5Ofmd6nWqq+/4wcHBb+/YseOtCRMmPOrp6flWbm7u6+fPn/fq2bPnVSeiwcHBr2/fvn3hhAkT/mYymV7Nz89/PCEhIWLcuHF/rG2fyZMn/05VVaOHh8dSnU6X5+Likl+ZrAYEBDx+/Pjxg6NGjfrQz8/vISFEQXl5+eiSkpLRW7dufbGuWFauXJkTERGRe/To0Un9+vX7BgB8fX1X7d279+9eXl4V69at2wgAHh4eSzp06PBGSkrKgZkzZ16/evXq0zNnzozOz8//0/bt2x+o3q6Pj88/y8rK7s7IyPg+Li7uxtLS0riEhIRROl3Dp9R1d3c/n5+fPyYuLs5z5cqVvLvGzruTdbhznQUd3q1AgAfwvym6WqduAoDufoB9JqrTcR5RImrbXF2doQ/QeTTkt+9tAIoBrAPgYlsuBlDZKxhhVwYAm2zLt9m1EW8rGwPgGdvy+wAgpUwGcD2ABwBkAUi2tTmnSWdUjy5duoz39PTM+vHHH3/dtWtXSocOHVbbJyFbtmx5ZcCAAf+Ij49/ZdOmTaVbtmzJj4+Pf8disQQAwIYNG74cM2bMIykpKbds3ry5YOvWrVnp6el/qulYPXv2fDgjIyNq/fr1Zfv27fs1KChoXX3x1Xf8zZs3vz148ODFhw4dennjxo0lWVlZ03v06JHUHD+bTZs2KcOHD3/j2LFjT2/YsKHs+PHjfx8xYsRrGzduXFTbPhaLJeD48eOvbdy4MW/Tpk1lGRkZQwYOHHgXAKxZs+bXESNGXJuenj5z27Zt2Rs3bjQfPHhwbWlpaZ+GxNOhQ4e9JSUl8Pb2/hgATCbT/8rKyhAaGppQWWfFihUVvXv3HghA99NPP50wGAzqTz/99Gtubu7kmtpctWpVyvDhwxecOXNm8oYNG4ri4+M/iIyM3OXi4lL/9QI2vr6+T5lMpoItW7bkG41Gdfr06WPq36t98PcQ+O5aFxQ+6IrUBa64pVfdXzFCCPS3m8nEOjRPRNR25ebmah2CUxGqWvsdrUTtwcCBA3fl5OT0SEpKCnTQIfghq8OzOyvw4k8q3FyAV0aW489DOT0tEbVdhYWF7eExn7UPhVXD/mFqdyZMmPCYwWBY6+rqeqSoqOju+Pj44f379/9I67jaq1FhAkY3FeUWoKCgAAATUSJquzIzM9tDItpgfNZ8GzJ69Oh3KidZr/6qPim+1rSMtaioaOTOnTsPrFu3znzw4MHXY2Nj1/n7+0tHHpNqNzREoNj2OAGO0BBRW9eQmWPaEw7NEzkeP2T1CF9oxm+FwH/GVOD+IbxznojaruLiYnh4tPmRnwYPzbNHlIg0NzpMwGwBLl7M1ToUIiKHOn/+vNYhOBUmokSkuYnh1j+eDQZDPTWJiFo3b29vrUNwKkxEiUhzw0MaPIpDRERtCBNRItJcjG3irN/yyrQNhIjIwfLy8rQOwakwESUizemEtUc0U/XROBIiIscKDQ3VOgSnwkSUiJzCjpt0+FPX37QOg4jIodLS0rQOwakwESUipzCqkw4hnOOZiNo4+8eKExNRInIigYGOesoqEZFz4PdcVUxEichpcMiKiNo6fs9VxUSUiJyGr6+v1iEQETkUv+eqYiJKRE7DbDZrHQIRkUPxe64qJqJE5DQKCgq0DoGIyKH4PVeVUFVV6xiI2jp+yBqotLQUer1e6zCIiBymnXzPNfhxeewRJSKncfbsWa1DICJyKH7PVcVElIichpubm9YhEBE5FL/nqmIiSkROw9/fX+sQiIgcit9zVTERJSKnkZ6ernUIREQOxe+5qly1DoCorRNClANwqW0zeDOTPa1+HnwfnAPfB+fA98E5tOb3oUxVVfeGVORd80QaEkLsU1V1sNZxOAshRKGqqi3+xHm+D86B74Nz4PvgHNrL+8CheSIiIiLSBBNRIiIiItIEE1EibSlaB+BkvtHouHwfnAPfB+fA98E5tIv3gdeIEhEREZEm2CNKRERERJpgIkpEREREmmAiSkRERESaYCJK1IyEEC5CiFeFEBeEEPlCiK+FEIF11J8uhDgqhCgWQhwRQkyttl0VQhQJIQrsXj6OP5PWrTHvgxAiTAjxvRAixfbzvrWGOlFCiI1CiEIhxFkhxCOOP4vWzwHvAz8PTdDI92GmEGKzECJTCJEjhNguhBhTrQ4/D03ggPehTXwemIgSNa/HAcwBMAxAJ1vZkpoqCiG6wXqX+MsAfGz/fiuE6FKt6lRVVU12r4sOibxtafD7AMACYD2AWwCcrb5RCOECYAWA4wCCAFwD4DEhxI3NHHNb1Gzvgx1+HhqvMe+DH4C3AETB+v99OYA1QojOAD8PV6nZ3gc7rf/zoKoqX3zx1UwvACkA/mC3HgnrI9q61FD3OQDbq5VtB/CM3boKYLTW59XaXo15H6rtlwzg1mplEwAUATDZlf0DwBatz9PZX835PtjK+XlowffBrv4FAHNty/w8OMH7YFtvE58H9ogSNRPbkEg4gP2VZaqqJgDIA9C3hl362de1OWArt/elbXhmjxDiumYMuU1qwvtQn34ATqqqWmBXVtP7RHYc8D5U4uehEa72fRBC9AUQAOCIrYifhyZwwPtQqdV/HpiIEjUfb9u/1YdGcu222fNqQN3JALrCOozzHwDLhBDTrzbQNq6x70N9GvI+0ZWa+30A+Hloiia/D0KIDgC+AvAvVVVP2Yr5eWia5n4fgDbyeWAiStR88m3/Vr9Y3BfWv3prql9nXVVVN6mqWmJ7fQ5gKYD5zRJt29XY96Eh7TVXW+1Jc78P/Dw0TZPeByFEKIAtsF63+0S19vh5aLzmfh/azOeBiShRM1FVNRdAKoCBlWW2G5K8ARyuYZdD9nVtBtjKa2MBIK4q0DauCe9DfQ4B6CGE8LQrq+99avcc8D7UhJ+HejTlfbDdMLkdwBpVVe9TbRck2vDz0AQOeB9q0io/D0xEiZqXAusdpF2FEN4A/glgnaqqyTXU/QTAYCHEzUIINyHEzQAGAVgMAEKI3kKIoUIId9v2awHcBuCLFjmT1q0x7wOEEAYhhAHWL3E327qrbfM2WG8yeEkI4SGE6A9gAYCFjj6JNqDZ3gd+Hq5Kg98HIUQ0gB0APlVV9dEa2uLnoema7X1oU58Hre+W4ouvtvQC4ALg3wAyYR2K+QZAoG3bfAAF1epPB3AUQLHt36l22ybYygoB5ADYB+Amrc+xNbya8D6oNbyetdseBWATrHcLnwfwqNbn2Bpezfk+8PPQMu8DgI9sP/eCaq/5dnX4edD4fWhLnwdhOyEiIiIiohbFoXkiIiIi0gQTUSIiIiLSBBNRIiIiItIEE1EiIiIi0gQTUSIiIiLSBBNRIiIiItKEa/1ViIiaTgihwjrf4Buqqv7drjwIwHIAg2GdTy8DQLqqqm80sN29AH6vqupRu7LNAEYC2Keq6ujmOgdFUfwBLAIwFdY5AJ+QUi5vSn1FUboAeBfACAClsD5D+kEppdm2fSuA4QDMtubOSSl7Nte5EBE5E/aIEtFVE0L4CSFUIURBtdcDtir97JNQmycAnFJV1Q/A4wBuh93TWYQQkUKIQiFEiF3ZfCHEeSFEZ1gnhn7evkFVVScCuNsBp/gOgDIAwbBOPP0/RVFim1j/XViT7hAA/QGMA3BPtf3vk1KabC8moUTUZjERJaLm0B9AtqqqpmqvN+vYZzKAL23LdwBYrapqceVGVVUTAKwE8CAACCFGAHgbwLWqqp4B8AOACfaJqiMoiuIJ4HoAT0kpC6SUO2zHvq2J9bsC+EJKWSKlTAOwFkBdSS0RUZvFoXkiag79ARxrSEUhhDuACwC8AawQQiTZ1j+sofo/AWwWQiyBdfj+blVV9wKAqqolQoj9sA5/L25MsIqirARQ29D9DillnN16DwAVUsqTdmWHYO3JrEl99d8EcJNtCN4PwAwAT1Vr42VFUV4BEA/g71LKrXWfERFR68RElIiawwA0MBFVVbXM1ru5RVXVYAAQQlyANemqXveA7VrQPQD+qarq59WqHAfQr7HBVks062MCcLFa2UUAXk2s/yOAuwDkwfrs6cUAvrOr+xisP8syADcBWKEoSn8pZUIjYiYiahWYiBJRc+gPIEoIcaNd2Yeqqj5cR/1Dduu+APKrVxJC6ABUALDA2jtaXT6s11o6UgGsvbf2vFFDvPXVVxRFB2AdrNfCjoQ1af0Q1nP7KwBIKffY7bdYUZSbAcwE8NZVnAMRkVNiIkpEV0UIoQfQC8AIVVX3NXC3/qiaiOag5h7G12BNUk/BetNP9eF7LwC5DY/WSlGUNQDG1LJ5u5Ryht36SQCuiqJ0l1KespX1A3D0yl3rre8PoDOAt6WUpQBKFUX5CMALsCWiNVABiIacFxFRayNUVdU6BiJqxYQQgwDsBuClqmppDdtVAN1VVT1tV7YOwBJVVZfa1jcC+EhV1WV2dRbAemf9MADjATwLIEa1+9ISQmwAsFRV1cV2ZXcA+GMzT9/0GawJ4R9hTaJXAxgppawxGa2rvqIoiQAUWO/6NwH4CECRlHK+oii+tvP9Edbpm2601R0opbzi0gUiotaOd80T0dUaAOBoTUloHfqhao/oatjd/COEmAzgJQCzVVVNh3WuTXcAc+zq6AEMArCh6aE32D0APGCddulTAH+yT0IVRVmjKMrfGlj/OgDTYb1B6zSsCedDtm1usPaOXoB1/tH7AVzLJJSI2ir2iBLRVRFCvA3r3J0ldsUqgE6qql6s3iMqhOgIIAWASVXVcltZIICDALoDiACwA8DtqqqutjvOvQBuVVV1hG19HoCbVVW9rlo8d6CZe0SJiMgxmIgSkUMJIUpgfYLQf1VVrT5NkX29lwBkNOLJSnsA/EFV1SN2ZRtgfSrRXlVVJ11V4ERE5HBMRImIiIhIE7xGlIiIiIg0wUSUiIiIiDTBRJSIiIiINMFElIiIiIg0wUSUiIiIiDTBRJSIiIiINMFElIiIiIg0wUSUiIiIiDTx/7ZfJ7jnIuf7AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 576x360 with 3 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "# Plot the SHAP values\n",
    "shap.plots.waterfall(shap_values[0])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This local plot (a waterfall plot) shows what are the main features affecting the prediction of a single observation, and the magnitude of the SHAP value for each feature. Here we can see how the sum of all SHAP values equals the difference between the prediction and the expected value."
   ]
  }
 ],
 "metadata": {
  "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": "3.9.7"
  },
  "vscode": {
   "interpreter": {
    "hash": "fc902f5f9f1b4ac8244fa2d1a71d10e1bb8b95bd909217946ac119cb82bcf206"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}