ps3e9-keras-mlp-selu-hyperopt.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "f72406be",
   "metadata": {
    "_cell_guid": "b1076dfc-b9ad-4769-8c92-a6c4dae69d19",
    "_uuid": "8f2839f25d086af736a60e9eeb907d3b93b6e0e5",
    "execution": {
     "iopub.execute_input": "2023-03-07T15:40:24.789314Z",
     "iopub.status.busy": "2023-03-07T15:40:24.788540Z",
     "iopub.status.idle": "2023-03-07T15:40:24.803801Z",
     "shell.execute_reply": "2023-03-07T15:40:24.802879Z"
    },
    "papermill": {
     "duration": 0.027925,
     "end_time": "2023-03-07T15:40:24.806606",
     "exception": false,
     "start_time": "2023-03-07T15:40:24.778681",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "/kaggle/input/playground-series-s3e9/sample_submission.csv\n",
      "/kaggle/input/playground-series-s3e9/train.csv\n",
      "/kaggle/input/playground-series-s3e9/test.csv\n"
     ]
    }
   ],
   "source": [
    "# This Python 3 environment comes with many helpful analytics libraries installed\n",
    "# It is defined by the kaggle/python Docker image: https://github.com/kaggle/docker-python\n",
    "# For example, here's several helpful packages to load\n",
    "\n",
    "import numpy as np # linear algebra\n",
    "import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)\n",
    "\n",
    "# Input data files are available in the read-only \"../input/\" directory\n",
    "# For example, running this (by clicking run or pressing Shift+Enter) will list all files under the input directory\n",
    "\n",
    "import os\n",
    "for dirname, _, filenames in os.walk('/kaggle/input'):\n",
    "    for filename in filenames:\n",
    "        print(os.path.join(dirname, filename))\n",
    "\n",
    "# You can write up to 20GB to the current directory (/kaggle/working/) that gets preserved as output when you create a version using \"Save & Run All\" \n",
    "# You can also write temporary files to /kaggle/temp/, but they won't be saved outside of the current session"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "bf2bb985",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-03-07T15:40:24.822818Z",
     "iopub.status.busy": "2023-03-07T15:40:24.822010Z",
     "iopub.status.idle": "2023-03-07T15:40:26.083743Z",
     "shell.execute_reply": "2023-03-07T15:40:26.082377Z"
    },
    "papermill": {
     "duration": 1.27278,
     "end_time": "2023-03-07T15:40:26.086904",
     "exception": false,
     "start_time": "2023-03-07T15:40:24.814124",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "308eab15",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-03-07T15:40:26.101688Z",
     "iopub.status.busy": "2023-03-07T15:40:26.101264Z",
     "iopub.status.idle": "2023-03-07T15:40:26.189602Z",
     "shell.execute_reply": "2023-03-07T15:40:26.188087Z"
    },
    "papermill": {
     "duration": 0.100198,
     "end_time": "2023-03-07T15:40:26.193443",
     "exception": false,
     "start_time": "2023-03-07T15:40:26.093245",
     "status": "completed"
    },
    "tags": []
   },
   "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>CementComponent</th>\n",
       "      <th>BlastFurnaceSlag</th>\n",
       "      <th>FlyAshComponent</th>\n",
       "      <th>WaterComponent</th>\n",
       "      <th>SuperplasticizerComponent</th>\n",
       "      <th>CoarseAggregateComponent</th>\n",
       "      <th>FineAggregateComponent</th>\n",
       "      <th>AgeInDays</th>\n",
       "      <th>Strength</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>525.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>186.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1125.0</td>\n",
       "      <td>613.0</td>\n",
       "      <td>3</td>\n",
       "      <td>10.38</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>143.0</td>\n",
       "      <td>169.0</td>\n",
       "      <td>143.0</td>\n",
       "      <td>191.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>967.0</td>\n",
       "      <td>643.0</td>\n",
       "      <td>28</td>\n",
       "      <td>23.52</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>289.0</td>\n",
       "      <td>134.7</td>\n",
       "      <td>0.0</td>\n",
       "      <td>185.7</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1075.0</td>\n",
       "      <td>795.3</td>\n",
       "      <td>28</td>\n",
       "      <td>36.96</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>304.0</td>\n",
       "      <td>76.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>228.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>932.0</td>\n",
       "      <td>670.0</td>\n",
       "      <td>365</td>\n",
       "      <td>39.05</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>157.0</td>\n",
       "      <td>236.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>192.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>935.4</td>\n",
       "      <td>781.2</td>\n",
       "      <td>90</td>\n",
       "      <td>74.19</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5402</th>\n",
       "      <td>446.0</td>\n",
       "      <td>24.0</td>\n",
       "      <td>79.0</td>\n",
       "      <td>162.0</td>\n",
       "      <td>11.6</td>\n",
       "      <td>967.0</td>\n",
       "      <td>712.0</td>\n",
       "      <td>3</td>\n",
       "      <td>15.42</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5403</th>\n",
       "      <td>350.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>203.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>974.0</td>\n",
       "      <td>775.0</td>\n",
       "      <td>180</td>\n",
       "      <td>49.20</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5404</th>\n",
       "      <td>295.8</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>185.7</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1076.2</td>\n",
       "      <td>759.3</td>\n",
       "      <td>28</td>\n",
       "      <td>39.30</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5405</th>\n",
       "      <td>376.0</td>\n",
       "      <td>93.4</td>\n",
       "      <td>0.0</td>\n",
       "      <td>162.6</td>\n",
       "      <td>11.5</td>\n",
       "      <td>955.8</td>\n",
       "      <td>662.9</td>\n",
       "      <td>28</td>\n",
       "      <td>39.61</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5406</th>\n",
       "      <td>190.7</td>\n",
       "      <td>0.0</td>\n",
       "      <td>125.2</td>\n",
       "      <td>166.6</td>\n",
       "      <td>7.9</td>\n",
       "      <td>1079.0</td>\n",
       "      <td>770.1</td>\n",
       "      <td>56</td>\n",
       "      <td>63.14</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5407 rows × 9 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      CementComponent  BlastFurnaceSlag  FlyAshComponent  WaterComponent  \\\n",
       "0               525.0               0.0              0.0           186.0   \n",
       "1               143.0             169.0            143.0           191.0   \n",
       "2               289.0             134.7              0.0           185.7   \n",
       "3               304.0              76.0              0.0           228.0   \n",
       "4               157.0             236.0              0.0           192.0   \n",
       "...               ...               ...              ...             ...   \n",
       "5402            446.0              24.0             79.0           162.0   \n",
       "5403            350.0               0.0              0.0           203.0   \n",
       "5404            295.8               0.0              0.0           185.7   \n",
       "5405            376.0              93.4              0.0           162.6   \n",
       "5406            190.7               0.0            125.2           166.6   \n",
       "\n",
       "      SuperplasticizerComponent  CoarseAggregateComponent  \\\n",
       "0                           0.0                    1125.0   \n",
       "1                           8.0                     967.0   \n",
       "2                           0.0                    1075.0   \n",
       "3                           0.0                     932.0   \n",
       "4                           0.0                     935.4   \n",
       "...                         ...                       ...   \n",
       "5402                       11.6                     967.0   \n",
       "5403                        0.0                     974.0   \n",
       "5404                        0.0                    1076.2   \n",
       "5405                       11.5                     955.8   \n",
       "5406                        7.9                    1079.0   \n",
       "\n",
       "      FineAggregateComponent  AgeInDays  Strength  \n",
       "0                      613.0          3     10.38  \n",
       "1                      643.0         28     23.52  \n",
       "2                      795.3         28     36.96  \n",
       "3                      670.0        365     39.05  \n",
       "4                      781.2         90     74.19  \n",
       "...                      ...        ...       ...  \n",
       "5402                   712.0          3     15.42  \n",
       "5403                   775.0        180     49.20  \n",
       "5404                   759.3         28     39.30  \n",
       "5405                   662.9         28     39.61  \n",
       "5406                   770.1         56     63.14  \n",
       "\n",
       "[5407 rows x 9 columns]"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train = pd.read_csv('/kaggle/input/playground-series-s3e9/train.csv')\n",
    "test = pd.read_csv('/kaggle/input/playground-series-s3e9/test.csv')\n",
    "del train['id']\n",
    "\n",
    "train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "7f56f303",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-03-07T15:40:26.208363Z",
     "iopub.status.busy": "2023-03-07T15:40:26.207922Z",
     "iopub.status.idle": "2023-03-07T15:40:26.230808Z",
     "shell.execute_reply": "2023-03-07T15:40:26.229617Z"
    },
    "papermill": {
     "duration": 0.033467,
     "end_time": "2023-03-07T15:40:26.233547",
     "exception": false,
     "start_time": "2023-03-07T15:40:26.200080",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "predictors = train.columns[:-1].to_list()\n",
    "train = train.groupby(predictors).agg('mean').reset_index()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "dfd0c3a4",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-03-07T15:40:26.249473Z",
     "iopub.status.busy": "2023-03-07T15:40:26.248867Z",
     "iopub.status.idle": "2023-03-07T15:40:26.600658Z",
     "shell.execute_reply": "2023-03-07T15:40:26.599289Z"
    },
    "papermill": {
     "duration": 0.363357,
     "end_time": "2023-03-07T15:40:26.603929",
     "exception": false,
     "start_time": "2023-03-07T15:40:26.240572",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "train['Strength'].hist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "2463e2fc",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-03-07T15:40:26.620263Z",
     "iopub.status.busy": "2023-03-07T15:40:26.619804Z",
     "iopub.status.idle": "2023-03-07T15:40:28.753003Z",
     "shell.execute_reply": "2023-03-07T15:40:28.751582Z"
    },
    "papermill": {
     "duration": 2.144694,
     "end_time": "2023-03-07T15:40:28.755766",
     "exception": false,
     "start_time": "2023-03-07T15:40:26.611072",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1600x800 with 8 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# histograms of predictors\n",
    "n_rows = 2\n",
    "n_cols = 4\n",
    "\n",
    "fig,axs = plt.subplots(n_rows,n_cols,figsize=(4*n_cols,4*n_rows))\n",
    "for i in range(n_rows):\n",
    "    for j in range(n_cols):\n",
    "        col_idx = n_cols*i+j\n",
    "        _ = sns.histplot(data=train,x=train.columns[col_idx],ax=axs[i][j])\n",
    "        \n",
    "fig.tight_layout()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a982ffb7",
   "metadata": {
    "papermill": {
     "duration": 0.008665,
     "end_time": "2023-03-07T15:40:28.773441",
     "exception": false,
     "start_time": "2023-03-07T15:40:28.764776",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Some of the predictors have a rather larger number of zero values. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "15afb184",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-03-07T15:40:28.793288Z",
     "iopub.status.busy": "2023-03-07T15:40:28.792499Z",
     "iopub.status.idle": "2023-03-07T15:40:28.801000Z",
     "shell.execute_reply": "2023-03-07T15:40:28.799546Z"
    },
    "papermill": {
     "duration": 0.021686,
     "end_time": "2023-03-07T15:40:28.803945",
     "exception": false,
     "start_time": "2023-03-07T15:40:28.782259",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Proportion of zero_values observations for col BlastFurnaceSlag: 0.537\n",
      "Proportion of zero_values observations for col FlyAshComponent: 0.623\n",
      "Proportion of zero_values observations for col SuperplasticizerComponent: 0.437\n"
     ]
    }
   ],
   "source": [
    "zero_value_cols = ['BlastFurnaceSlag','FlyAshComponent','SuperplasticizerComponent']\n",
    "for col in zero_value_cols:\n",
    "    print('Proportion of zero_values observations for col %s: %5.3f' %(col,(train[col]==0).mean()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "89e99233",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-03-07T15:40:28.824422Z",
     "iopub.status.busy": "2023-03-07T15:40:28.823624Z",
     "iopub.status.idle": "2023-03-07T15:40:29.548229Z",
     "shell.execute_reply": "2023-03-07T15:40:29.546719Z"
    },
    "papermill": {
     "duration": 0.738782,
     "end_time": "2023-03-07T15:40:29.551358",
     "exception": false,
     "start_time": "2023-03-07T15:40:28.812576",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 900x400 with 3 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig,axs = plt.subplots(1,len(zero_value_cols),figsize=(3*len(zero_value_cols),4))\n",
    "for i,col in enumerate(zero_value_cols):\n",
    "    indicator = 1*(train[col] == 0)\n",
    "    _ = sns.boxplot(y=train['Strength'],x=indicator,ax=axs[i])\n",
    "\n",
    "fig.tight_layout()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "4f267c27",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-03-07T15:40:29.572377Z",
     "iopub.status.busy": "2023-03-07T15:40:29.571401Z",
     "iopub.status.idle": "2023-03-07T15:40:29.761790Z",
     "shell.execute_reply": "2023-03-07T15:40:29.760442Z"
    },
    "papermill": {
     "duration": 0.203988,
     "end_time": "2023-03-07T15:40:29.764954",
     "exception": false,
     "start_time": "2023-03-07T15:40:29.560966",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import StandardScaler\n",
    "\n",
    "\n",
    "input_sc = StandardScaler()\n",
    "output_sc = StandardScaler()\n",
    "\n",
    "X = input_sc.fit_transform(train.drop('Strength',axis=1).values)\n",
    "y = output_sc.fit_transform(train[['Strength']].values).ravel()\n",
    "original_std = output_sc.scale_"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0c76ec51",
   "metadata": {
    "papermill": {
     "duration": 0.008972,
     "end_time": "2023-03-07T15:40:29.783032",
     "exception": false,
     "start_time": "2023-03-07T15:40:29.774060",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## 2-layer multi-layer perceptron"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "4876e992",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-03-07T15:40:29.805557Z",
     "iopub.status.busy": "2023-03-07T15:40:29.803988Z",
     "iopub.status.idle": "2023-03-07T15:40:39.756295Z",
     "shell.execute_reply": "2023-03-07T15:40:39.754789Z"
    },
    "papermill": {
     "duration": 9.965806,
     "end_time": "2023-03-07T15:40:39.759416",
     "exception": false,
     "start_time": "2023-03-07T15:40:29.793610",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "import keras\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "from typing import Dict, Optional"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "5c3dd589",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-03-07T15:40:39.779814Z",
     "iopub.status.busy": "2023-03-07T15:40:39.778810Z",
     "iopub.status.idle": "2023-03-07T15:40:39.794802Z",
     "shell.execute_reply": "2023-03-07T15:40:39.793144Z"
    },
    "papermill": {
     "duration": 0.029128,
     "end_time": "2023-03-07T15:40:39.797391",
     "exception": false,
     "start_time": "2023-03-07T15:40:39.768263",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "def construct_model(params:Dict) -> keras.Model:\n",
    "    \"\"\"\n",
    "    Build 2-layer MLP for hyperparameters tuning\n",
    "\n",
    "    hp: HyperParameters class instance\n",
    "    \"\"\"\n",
    "    n_hidden = 2 # TODO: use this as a hyperparameter\n",
    "    n_hidden_list = [params['hsize%d'%i] for i in range(1,n_hidden+1)]\n",
    "    dropouts_list = [params['dropout%d'%i] for i in range(1,n_hidden+1)]\n",
    "\n",
    "    model = keras.Sequential()\n",
    "    model.add(keras.layers.Dense(n_hidden_list[0],input_shape=(X.shape[-1],),activation='selu',kernel_initializer='lecun_normal'))\n",
    "    model.add(keras.layers.AlphaDropout(dropouts_list[0]))\n",
    "\n",
    "    if n_hidden > 1:\n",
    "        for i in range(1,n_hidden):\n",
    "            model.add(keras.layers.Dense(n_hidden_list[i],activation='selu',kernel_initializer='lecun_normal'))\n",
    "            model.add(keras.layers.AlphaDropout(dropouts_list[i]))\n",
    "            \n",
    "    # output is between 0 and 1\n",
    "    model.add(keras.layers.Dense(1))\n",
    "\n",
    "    model.compile(\n",
    "        optimizer=keras.optimizers.Adam(learning_rate=params['learning_rate']),\n",
    "        loss='mean_squared_error'\n",
    "    )\n",
    "\n",
    "    return model\n",
    "\n",
    "\n",
    "def build_and_fit(\n",
    "    params:Dict,X:np.ndarray,y:np.ndarray,\n",
    "    verbose:bool=False,\n",
    "    resample_seed:Optional[int]=None\n",
    ") -> keras.Model:\n",
    "    #further divide into train and validation splits for early stopping\n",
    "    X_train,X_val,y_train,y_val = train_test_split(X,y,test_size=0.1,random_state=resample_seed)\n",
    "    \n",
    "    EPOCHS = 100\n",
    "    BATCH_SIZE = params['batch_size']\n",
    "        \n",
    "    early_stopping = keras.callbacks.EarlyStopping(\n",
    "        monitor='val_loss',\n",
    "        mode='min',\n",
    "        verbose=0,\n",
    "        patience=10,\n",
    "        restore_best_weights=True\n",
    "    )\n",
    "        \n",
    "        \n",
    "    # create model\n",
    "    model = construct_model(params)\n",
    "\n",
    "    # train model\n",
    "    _ = model.fit(\n",
    "        X_train,\n",
    "        y_train,\n",
    "        batch_size=BATCH_SIZE,\n",
    "        epochs=EPOCHS,\n",
    "        callbacks=[early_stopping],\n",
    "        validation_data=(X_val, y_val),\n",
    "        verbose=verbose,\n",
    "    )\n",
    "\n",
    "    return model"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "799b679e",
   "metadata": {
    "papermill": {
     "duration": 0.008533,
     "end_time": "2023-03-07T15:40:39.815024",
     "exception": false,
     "start_time": "2023-03-07T15:40:39.806491",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "### Hyperparameter tuning\n",
    "\n",
    "We will use the hyperopt library to tune the hyperparameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "7e65ad71",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-03-07T15:40:39.835732Z",
     "iopub.status.busy": "2023-03-07T15:40:39.834920Z",
     "iopub.status.idle": "2023-03-07T15:40:40.697959Z",
     "shell.execute_reply": "2023-03-07T15:40:40.695700Z"
    },
    "papermill": {
     "duration": 0.879386,
     "end_time": "2023-03-07T15:40:40.703289",
     "exception": false,
     "start_time": "2023-03-07T15:40:39.823903",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "from sklearn.model_selection import KFold\n",
    "from sklearn.metrics import mean_squared_error\n",
    "from joblib import Parallel, delayed\n",
    "\n",
    "from hyperopt import STATUS_OK, Trials, fmin, hp, tpe\n",
    "from hyperopt.pyll_utils import scope"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "fca83d84",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-03-07T15:40:40.723926Z",
     "iopub.status.busy": "2023-03-07T15:40:40.723486Z",
     "iopub.status.idle": "2023-03-07T15:40:40.736611Z",
     "shell.execute_reply": "2023-03-07T15:40:40.735258Z"
    },
    "papermill": {
     "duration": 0.027321,
     "end_time": "2023-03-07T15:40:40.739685",
     "exception": false,
     "start_time": "2023-03-07T15:40:40.712364",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "8"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# generate train_test_splits\n",
    "splits = list(KFold(n_splits=8,shuffle=True,random_state=1).split(X,y))\n",
    "len(splits)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "f2069476",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-03-07T15:40:40.760283Z",
     "iopub.status.busy": "2023-03-07T15:40:40.759522Z",
     "iopub.status.idle": "2023-03-07T16:02:47.890261Z",
     "shell.execute_reply": "2023-03-07T16:02:47.889173Z"
    },
    "papermill": {
     "duration": 1327.14455,
     "end_time": "2023-03-07T16:02:47.893401",
     "exception": false,
     "start_time": "2023-03-07T15:40:40.748851",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " 64%|██████▍   | 64/100 [14:06<05:40,  9.45s/trial, best loss: 12.079809119124455]"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/conda/lib/python3.7/site-packages/joblib/externals/loky/process_executor.py:703: UserWarning: A worker stopped while some jobs were given to the executor. This can be caused by a too short worker timeout or by a memory leak.\n",
      "  \"timeout or by a memory leak.\", UserWarning\n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "100%|██████████| 100/100 [22:07<00:00, 13.27s/trial, best loss: 12.037573064462928]\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'batch_size': 430,\n",
       " 'dropout1': 0.010450274259104476,\n",
       " 'dropout2': 0.03438848058867812,\n",
       " 'hsize1': 46,\n",
       " 'hsize2': 128,\n",
       " 'learning_rate': 0.0077311904033691295}"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def fit_and_test_fold(params:Dict,train_index:np.ndarray,test_index:np.ndarray) -> float:\n",
    "    X_train = X[train_index,...]; X_test = X[test_index,...]\n",
    "    y_train = y[train_index]; y_test = y[test_index]\n",
    "    \n",
    "    \n",
    "    model = build_and_fit(params,X_train,y_train)\n",
    "\n",
    "    # obtain test predictions\n",
    "    test_preds = model.predict(X_test,verbose=False).ravel()\n",
    "    \n",
    "    # return RMSE\n",
    "    return np.sqrt(mean_squared_error(y_test,test_preds))*original_std\n",
    "\n",
    "def objective(params:Dict) -> Dict:\n",
    "    tf.keras.backend.clear_session()\n",
    "#     cv_losses = [None]*len(splits)\n",
    "#     for i,(train_index,test_index) in enumerate(splits):\n",
    "#         cv_losses[i] = fit_and_test_fold(params,train_index,test_index)\n",
    "    \n",
    "    # NOT using GPU in this notebook\n",
    "    cv_losses = Parallel(n_jobs=4,verbose=0)(\n",
    "        delayed(fit_and_test_fold)(params,train_index,test_index) for train_index,test_index in splits\n",
    "    )\n",
    "    \n",
    "    return {'loss':np.mean(cv_losses),'status': STATUS_OK,'min_RMSE_fold':np.min(cv_losses),'max_RMSE_fold':np.max(cv_losses)}\n",
    "\n",
    "search_space = {\n",
    "    'hsize1': scope.int(hp.quniform('hsize1',2,150,1)),\n",
    "    'hsize2': scope.int(hp.quniform('hsize2',2,150,1)),\n",
    "    'dropout1': hp.uniform('dropout1',0.01,0.5),\n",
    "    'dropout2': hp.uniform('dropout2',0.01,0.5),\n",
    "    'learning_rate': hp.loguniform('learning_rate',np.log(1e-4),np.log(0.1)),\n",
    "    'batch_size':scope.int(hp.quniform('batch_size',32,1024,1)),\n",
    "}\n",
    "\n",
    "trials = Trials()\n",
    "best = fmin(\n",
    "    objective,search_space,\n",
    "    algo=tpe.suggest,\n",
    "    max_evals = 100,\n",
    "    trials=trials,\n",
    "    return_argmin=False,\n",
    "    rstate=np.random.default_rng(1)\n",
    ")\n",
    "best"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "cace4e3a",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-03-07T16:02:47.928988Z",
     "iopub.status.busy": "2023-03-07T16:02:47.928091Z",
     "iopub.status.idle": "2023-03-07T16:02:47.967556Z",
     "shell.execute_reply": "2023-03-07T16:02:47.966009Z"
    },
    "papermill": {
     "duration": 0.060953,
     "end_time": "2023-03-07T16:02:47.970523",
     "exception": false,
     "start_time": "2023-03-07T16:02:47.909570",
     "status": "completed"
    },
    "tags": []
   },
   "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>batch_size</th>\n",
       "      <th>dropout1</th>\n",
       "      <th>dropout2</th>\n",
       "      <th>hsize1</th>\n",
       "      <th>hsize2</th>\n",
       "      <th>learning_rate</th>\n",
       "      <th>RMSE</th>\n",
       "      <th>min_RMSE_fold</th>\n",
       "      <th>max_RMSE_fold</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>67</th>\n",
       "      <td>430.0</td>\n",
       "      <td>0.010450</td>\n",
       "      <td>0.034388</td>\n",
       "      <td>46.0</td>\n",
       "      <td>128.0</td>\n",
       "      <td>0.007731</td>\n",
       "      <td>12.037573</td>\n",
       "      <td>11.285827</td>\n",
       "      <td>12.800012</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>68</th>\n",
       "      <td>537.0</td>\n",
       "      <td>0.022843</td>\n",
       "      <td>0.034701</td>\n",
       "      <td>54.0</td>\n",
       "      <td>130.0</td>\n",
       "      <td>0.006784</td>\n",
       "      <td>12.046509</td>\n",
       "      <td>11.352854</td>\n",
       "      <td>12.796913</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>65</th>\n",
       "      <td>494.0</td>\n",
       "      <td>0.022972</td>\n",
       "      <td>0.040263</td>\n",
       "      <td>20.0</td>\n",
       "      <td>118.0</td>\n",
       "      <td>0.016112</td>\n",
       "      <td>12.048721</td>\n",
       "      <td>11.445551</td>\n",
       "      <td>12.823399</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>66</th>\n",
       "      <td>483.0</td>\n",
       "      <td>0.019805</td>\n",
       "      <td>0.039815</td>\n",
       "      <td>45.0</td>\n",
       "      <td>128.0</td>\n",
       "      <td>0.017562</td>\n",
       "      <td>12.054181</td>\n",
       "      <td>11.451809</td>\n",
       "      <td>12.757066</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>86</th>\n",
       "      <td>422.0</td>\n",
       "      <td>0.019448</td>\n",
       "      <td>0.024650</td>\n",
       "      <td>34.0</td>\n",
       "      <td>113.0</td>\n",
       "      <td>0.005445</td>\n",
       "      <td>12.054932</td>\n",
       "      <td>11.321574</td>\n",
       "      <td>12.865140</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>71</th>\n",
       "      <td>431.0</td>\n",
       "      <td>0.040221</td>\n",
       "      <td>0.085420</td>\n",
       "      <td>69.0</td>\n",
       "      <td>124.0</td>\n",
       "      <td>0.003436</td>\n",
       "      <td>12.059578</td>\n",
       "      <td>11.287325</td>\n",
       "      <td>12.849941</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73</th>\n",
       "      <td>605.0</td>\n",
       "      <td>0.058600</td>\n",
       "      <td>0.057884</td>\n",
       "      <td>23.0</td>\n",
       "      <td>140.0</td>\n",
       "      <td>0.032501</td>\n",
       "      <td>12.065927</td>\n",
       "      <td>11.346157</td>\n",
       "      <td>13.221272</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>57</th>\n",
       "      <td>409.0</td>\n",
       "      <td>0.011010</td>\n",
       "      <td>0.014629</td>\n",
       "      <td>42.0</td>\n",
       "      <td>109.0</td>\n",
       "      <td>0.010106</td>\n",
       "      <td>12.079809</td>\n",
       "      <td>11.445714</td>\n",
       "      <td>12.827628</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>78</th>\n",
       "      <td>367.0</td>\n",
       "      <td>0.067836</td>\n",
       "      <td>0.033089</td>\n",
       "      <td>45.0</td>\n",
       "      <td>103.0</td>\n",
       "      <td>0.020363</td>\n",
       "      <td>12.093362</td>\n",
       "      <td>11.353747</td>\n",
       "      <td>12.952374</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>79</th>\n",
       "      <td>561.0</td>\n",
       "      <td>0.030116</td>\n",
       "      <td>0.054389</td>\n",
       "      <td>56.0</td>\n",
       "      <td>113.0</td>\n",
       "      <td>0.004165</td>\n",
       "      <td>12.131860</td>\n",
       "      <td>11.564171</td>\n",
       "      <td>12.851304</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    batch_size  dropout1  dropout2  hsize1  hsize2  learning_rate       RMSE  \\\n",
       "67       430.0  0.010450  0.034388    46.0   128.0       0.007731  12.037573   \n",
       "68       537.0  0.022843  0.034701    54.0   130.0       0.006784  12.046509   \n",
       "65       494.0  0.022972  0.040263    20.0   118.0       0.016112  12.048721   \n",
       "66       483.0  0.019805  0.039815    45.0   128.0       0.017562  12.054181   \n",
       "86       422.0  0.019448  0.024650    34.0   113.0       0.005445  12.054932   \n",
       "71       431.0  0.040221  0.085420    69.0   124.0       0.003436  12.059578   \n",
       "73       605.0  0.058600  0.057884    23.0   140.0       0.032501  12.065927   \n",
       "57       409.0  0.011010  0.014629    42.0   109.0       0.010106  12.079809   \n",
       "78       367.0  0.067836  0.033089    45.0   103.0       0.020363  12.093362   \n",
       "79       561.0  0.030116  0.054389    56.0   113.0       0.004165  12.131860   \n",
       "\n",
       "    min_RMSE_fold  max_RMSE_fold  \n",
       "67      11.285827      12.800012  \n",
       "68      11.352854      12.796913  \n",
       "65      11.445551      12.823399  \n",
       "66      11.451809      12.757066  \n",
       "86      11.321574      12.865140  \n",
       "71      11.287325      12.849941  \n",
       "73      11.346157      13.221272  \n",
       "57      11.445714      12.827628  \n",
       "78      11.353747      12.952374  \n",
       "79      11.564171      12.851304  "
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "results = pd.DataFrame(trials.vals)\n",
    "results['RMSE'] = [tmp['loss'] for tmp in trials.results]\n",
    "results['min_RMSE_fold'] = [tmp['min_RMSE_fold'] for tmp in trials.results]\n",
    "results['max_RMSE_fold'] = [tmp['max_RMSE_fold'] for tmp in trials.results]\n",
    "results.to_csv('cv_rmse_hyperopt.csv',index=False)\n",
    "# top 10 configurations\n",
    "results.sort_values('RMSE',ascending=True).head(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bb960d4c",
   "metadata": {
    "papermill": {
     "duration": 0.016937,
     "end_time": "2023-03-07T16:02:48.004904",
     "exception": false,
     "start_time": "2023-03-07T16:02:47.987967",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Final model\n",
    "\n",
    "Rather than training a single model, we will train an ensemble of models, where each model is trained using a different training-validation split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "304812af",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-03-07T16:02:48.041885Z",
     "iopub.status.busy": "2023-03-07T16:02:48.041412Z",
     "iopub.status.idle": "2023-03-07T16:03:29.302804Z",
     "shell.execute_reply": "2023-03-07T16:03:29.301540Z"
    },
    "papermill": {
     "duration": 41.283561,
     "end_time": "2023-03-07T16:03:29.305259",
     "exception": false,
     "start_time": "2023-03-07T16:02:48.021698",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "********* Model 1 *********\n",
      "Epoch 1/100\n",
      "7/7 [==============================] - 1s 43ms/step - loss: 1.6436 - val_loss: 1.1069\n",
      "Epoch 2/100\n",
      "7/7 [==============================] - 0s 11ms/step - loss: 1.1401 - val_loss: 0.8464\n",
      "Epoch 3/100\n",
      "7/7 [==============================] - 0s 11ms/step - loss: 0.8935 - val_loss: 0.7811\n",
      "Epoch 4/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.8425 - val_loss: 0.7297\n",
      "Epoch 5/100\n",
      "7/7 [==============================] - 0s 12ms/step - loss: 0.7716 - val_loss: 0.7476\n",
      "Epoch 6/100\n",
      "7/7 [==============================] - 0s 12ms/step - loss: 0.7414 - val_loss: 0.6818\n",
      "Epoch 7/100\n",
      "7/7 [==============================] - 0s 11ms/step - loss: 0.7276 - val_loss: 0.6958\n",
      "Epoch 8/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.7091 - val_loss: 0.6946\n",
      "Epoch 9/100\n",
      "7/7 [==============================] - 0s 11ms/step - loss: 0.6853 - val_loss: 0.6691\n",
      "Epoch 10/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6744 - val_loss: 0.6565\n",
      "Epoch 11/100\n",
      "7/7 [==============================] - 0s 14ms/step - loss: 0.6577 - val_loss: 0.6527\n",
      "Epoch 12/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6425 - val_loss: 0.6851\n",
      "Epoch 13/100\n",
      "7/7 [==============================] - 0s 11ms/step - loss: 0.6524 - val_loss: 0.6293\n",
      "Epoch 14/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6379 - val_loss: 0.6436\n",
      "Epoch 15/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6336 - val_loss: 0.6248\n",
      "Epoch 16/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6309 - val_loss: 0.6308\n",
      "Epoch 17/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6137 - val_loss: 0.6314\n",
      "Epoch 18/100\n",
      "7/7 [==============================] - 0s 11ms/step - loss: 0.6250 - val_loss: 0.5986\n",
      "Epoch 19/100\n",
      "7/7 [==============================] - 0s 11ms/step - loss: 0.5925 - val_loss: 0.6105\n",
      "Epoch 20/100\n",
      "7/7 [==============================] - 0s 11ms/step - loss: 0.5886 - val_loss: 0.5850\n",
      "Epoch 21/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5828 - val_loss: 0.6076\n",
      "Epoch 22/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5894 - val_loss: 0.6099\n",
      "Epoch 23/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5814 - val_loss: 0.5755\n",
      "Epoch 24/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5714 - val_loss: 0.5852\n",
      "Epoch 25/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5667 - val_loss: 0.5821\n",
      "Epoch 26/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5668 - val_loss: 0.5873\n",
      "Epoch 27/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5699 - val_loss: 0.5902\n",
      "Epoch 28/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5702 - val_loss: 0.6038\n",
      "Epoch 29/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5713 - val_loss: 0.5845\n",
      "Epoch 30/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5697 - val_loss: 0.5805\n",
      "Epoch 31/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5531 - val_loss: 0.5836\n",
      "Epoch 32/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5667 - val_loss: 0.5938\n",
      "Epoch 33/100\n",
      "7/7 [==============================] - 0s 11ms/step - loss: 0.5648 - val_loss: 0.5812\n",
      "\n",
      "********* Model 2 *********\n",
      "Epoch 1/100\n",
      "7/7 [==============================] - 1s 31ms/step - loss: 1.9617 - val_loss: 1.3177\n",
      "Epoch 2/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 1.1978 - val_loss: 0.9410\n",
      "Epoch 3/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.9646 - val_loss: 0.7353\n",
      "Epoch 4/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.8625 - val_loss: 0.7508\n",
      "Epoch 5/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.7842 - val_loss: 0.6496\n",
      "Epoch 6/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.7898 - val_loss: 0.7156\n",
      "Epoch 7/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.7381 - val_loss: 0.6651\n",
      "Epoch 8/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.7449 - val_loss: 0.6605\n",
      "Epoch 9/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6952 - val_loss: 0.6527\n",
      "Epoch 10/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.7092 - val_loss: 0.6856\n",
      "Epoch 11/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6958 - val_loss: 0.6194\n",
      "Epoch 12/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6644 - val_loss: 0.6561\n",
      "Epoch 13/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6963 - val_loss: 0.6565\n",
      "Epoch 14/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6594 - val_loss: 0.6408\n",
      "Epoch 15/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6340 - val_loss: 0.6333\n",
      "Epoch 16/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6196 - val_loss: 0.6198\n",
      "Epoch 17/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6170 - val_loss: 0.6161\n",
      "Epoch 18/100\n",
      "7/7 [==============================] - 0s 13ms/step - loss: 0.6042 - val_loss: 0.6187\n",
      "Epoch 19/100\n",
      "7/7 [==============================] - 0s 11ms/step - loss: 0.6012 - val_loss: 0.6086\n",
      "Epoch 20/100\n",
      "7/7 [==============================] - 0s 11ms/step - loss: 0.6091 - val_loss: 0.6126\n",
      "Epoch 21/100\n",
      "7/7 [==============================] - 0s 13ms/step - loss: 0.5970 - val_loss: 0.6050\n",
      "Epoch 22/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5772 - val_loss: 0.6089\n",
      "Epoch 23/100\n",
      "7/7 [==============================] - 0s 11ms/step - loss: 0.5813 - val_loss: 0.5929\n",
      "Epoch 24/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5800 - val_loss: 0.5933\n",
      "Epoch 25/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5792 - val_loss: 0.5779\n",
      "Epoch 26/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5731 - val_loss: 0.5858\n",
      "Epoch 27/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5804 - val_loss: 0.6248\n",
      "Epoch 28/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5888 - val_loss: 0.5881\n",
      "Epoch 29/100\n",
      "7/7 [==============================] - 0s 12ms/step - loss: 0.5775 - val_loss: 0.6010\n",
      "Epoch 30/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5815 - val_loss: 0.6168\n",
      "Epoch 31/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5654 - val_loss: 0.5945\n",
      "Epoch 32/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5701 - val_loss: 0.5840\n",
      "Epoch 33/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5676 - val_loss: 0.5814\n",
      "Epoch 34/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5649 - val_loss: 0.5894\n",
      "Epoch 35/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5602 - val_loss: 0.5770\n",
      "Epoch 36/100\n",
      "7/7 [==============================] - 0s 11ms/step - loss: 0.5642 - val_loss: 0.5924\n",
      "Epoch 37/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5620 - val_loss: 0.5885\n",
      "Epoch 38/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5642 - val_loss: 0.5918\n",
      "Epoch 39/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5623 - val_loss: 0.5895\n",
      "Epoch 40/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5567 - val_loss: 0.5922\n",
      "Epoch 41/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5573 - val_loss: 0.5729\n",
      "Epoch 42/100\n",
      "7/7 [==============================] - 0s 13ms/step - loss: 0.5602 - val_loss: 0.6116\n",
      "Epoch 43/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5568 - val_loss: 0.5913\n",
      "Epoch 44/100\n",
      "7/7 [==============================] - 0s 13ms/step - loss: 0.5677 - val_loss: 0.6085\n",
      "Epoch 45/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5630 - val_loss: 0.5811\n",
      "Epoch 46/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5553 - val_loss: 0.5915\n",
      "Epoch 47/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5512 - val_loss: 0.5857\n",
      "Epoch 48/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5517 - val_loss: 0.5962\n",
      "Epoch 49/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5570 - val_loss: 0.6038\n",
      "Epoch 50/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5517 - val_loss: 0.5936\n",
      "Epoch 51/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5525 - val_loss: 0.5836\n",
      "\n",
      "********* Model 3 *********\n",
      "Epoch 1/100\n",
      "7/7 [==============================] - 1s 31ms/step - loss: 2.2841 - val_loss: 1.4481\n",
      "Epoch 2/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 1.2908 - val_loss: 1.1216\n",
      "Epoch 3/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 1.0599 - val_loss: 0.8927\n",
      "Epoch 4/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.9814 - val_loss: 0.8383\n",
      "Epoch 5/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.8782 - val_loss: 0.8054\n",
      "Epoch 6/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.8595 - val_loss: 0.7875\n",
      "Epoch 7/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.7946 - val_loss: 0.7468\n",
      "Epoch 8/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.7610 - val_loss: 0.7722\n",
      "Epoch 9/100\n",
      "7/7 [==============================] - 0s 13ms/step - loss: 0.7303 - val_loss: 0.7266\n",
      "Epoch 10/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.7237 - val_loss: 0.7362\n",
      "Epoch 11/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.7184 - val_loss: 0.7165\n",
      "Epoch 12/100\n",
      "7/7 [==============================] - 0s 11ms/step - loss: 0.6999 - val_loss: 0.6999\n",
      "Epoch 13/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6869 - val_loss: 0.7254\n",
      "Epoch 14/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6801 - val_loss: 0.6861\n",
      "Epoch 15/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.6679 - val_loss: 0.7004\n",
      "Epoch 16/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6752 - val_loss: 0.6826\n",
      "Epoch 17/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6405 - val_loss: 0.6749\n",
      "Epoch 18/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6475 - val_loss: 0.6518\n",
      "Epoch 19/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6284 - val_loss: 0.6630\n",
      "Epoch 20/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6011 - val_loss: 0.6437\n",
      "Epoch 21/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6084 - val_loss: 0.6388\n",
      "Epoch 22/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.6038 - val_loss: 0.6338\n",
      "Epoch 23/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5893 - val_loss: 0.6120\n",
      "Epoch 24/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5781 - val_loss: 0.6188\n",
      "Epoch 25/100\n",
      "7/7 [==============================] - 0s 13ms/step - loss: 0.5884 - val_loss: 0.6107\n",
      "Epoch 26/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5744 - val_loss: 0.6085\n",
      "Epoch 27/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5694 - val_loss: 0.6131\n",
      "Epoch 28/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5765 - val_loss: 0.6099\n",
      "Epoch 29/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5763 - val_loss: 0.6037\n",
      "Epoch 30/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5789 - val_loss: 0.6158\n",
      "Epoch 31/100\n",
      "7/7 [==============================] - 0s 11ms/step - loss: 0.5699 - val_loss: 0.5992\n",
      "Epoch 32/100\n",
      "7/7 [==============================] - 0s 12ms/step - loss: 0.5676 - val_loss: 0.6190\n",
      "Epoch 33/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5818 - val_loss: 0.6030\n",
      "Epoch 34/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5557 - val_loss: 0.6265\n",
      "Epoch 35/100\n",
      "7/7 [==============================] - 0s 11ms/step - loss: 0.5656 - val_loss: 0.6086\n",
      "Epoch 36/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5644 - val_loss: 0.6259\n",
      "Epoch 37/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5538 - val_loss: 0.6032\n",
      "Epoch 38/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5606 - val_loss: 0.6043\n",
      "Epoch 39/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5605 - val_loss: 0.6106\n",
      "Epoch 40/100\n",
      "7/7 [==============================] - 0s 11ms/step - loss: 0.5611 - val_loss: 0.5961\n",
      "Epoch 41/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5608 - val_loss: 0.6093\n",
      "Epoch 42/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5602 - val_loss: 0.6030\n",
      "Epoch 43/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5553 - val_loss: 0.6050\n",
      "Epoch 44/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5429 - val_loss: 0.6011\n",
      "Epoch 45/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5544 - val_loss: 0.6098\n",
      "Epoch 46/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5554 - val_loss: 0.6235\n",
      "Epoch 47/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5595 - val_loss: 0.6139\n",
      "Epoch 48/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5588 - val_loss: 0.6011\n",
      "Epoch 49/100\n",
      "7/7 [==============================] - 0s 13ms/step - loss: 0.5482 - val_loss: 0.6063\n",
      "Epoch 50/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5556 - val_loss: 0.6133\n",
      "\n",
      "********* Model 4 *********\n",
      "Epoch 1/100\n",
      "7/7 [==============================] - 1s 75ms/step - loss: 1.9807 - val_loss: 1.4090\n",
      "Epoch 2/100\n",
      "7/7 [==============================] - 0s 11ms/step - loss: 1.2772 - val_loss: 1.0023\n",
      "Epoch 3/100\n",
      "7/7 [==============================] - 0s 11ms/step - loss: 1.0519 - val_loss: 0.8103\n",
      "Epoch 4/100\n",
      "7/7 [==============================] - 0s 11ms/step - loss: 0.9204 - val_loss: 0.8003\n",
      "Epoch 5/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.8721 - val_loss: 0.7545\n",
      "Epoch 6/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.7816 - val_loss: 0.7114\n",
      "Epoch 7/100\n",
      "7/7 [==============================] - 0s 11ms/step - loss: 0.7742 - val_loss: 0.7101\n",
      "Epoch 8/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.7479 - val_loss: 0.7253\n",
      "Epoch 9/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.7235 - val_loss: 0.6859\n",
      "Epoch 10/100\n",
      "7/7 [==============================] - 0s 13ms/step - loss: 0.7217 - val_loss: 0.7027\n",
      "Epoch 11/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.7166 - val_loss: 0.7081\n",
      "Epoch 12/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.7028 - val_loss: 0.7073\n",
      "Epoch 13/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6762 - val_loss: 0.6820\n",
      "Epoch 14/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6660 - val_loss: 0.6806\n",
      "Epoch 15/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6471 - val_loss: 0.6976\n",
      "Epoch 16/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6649 - val_loss: 0.6670\n",
      "Epoch 17/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6546 - val_loss: 0.7095\n",
      "Epoch 18/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6548 - val_loss: 0.6691\n",
      "Epoch 19/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6218 - val_loss: 0.6673\n",
      "Epoch 20/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6222 - val_loss: 0.6293\n",
      "Epoch 21/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.6177 - val_loss: 0.6511\n",
      "Epoch 22/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6015 - val_loss: 0.6308\n",
      "Epoch 23/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5974 - val_loss: 0.6351\n",
      "Epoch 24/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5906 - val_loss: 0.6164\n",
      "Epoch 25/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5804 - val_loss: 0.6395\n",
      "Epoch 26/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5907 - val_loss: 0.6484\n",
      "Epoch 27/100\n",
      "7/7 [==============================] - 0s 11ms/step - loss: 0.5864 - val_loss: 0.6304\n",
      "Epoch 28/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5830 - val_loss: 0.6273\n",
      "Epoch 29/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5782 - val_loss: 0.6227\n",
      "Epoch 30/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5688 - val_loss: 0.6255\n",
      "Epoch 31/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5656 - val_loss: 0.6241\n",
      "Epoch 32/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5817 - val_loss: 0.6289\n",
      "Epoch 33/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5680 - val_loss: 0.6226\n",
      "Epoch 34/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5650 - val_loss: 0.6307\n",
      "\n",
      "********* Model 5 *********\n",
      "Epoch 1/100\n",
      "7/7 [==============================] - 1s 29ms/step - loss: 1.9863 - val_loss: 1.1131\n",
      "Epoch 2/100\n",
      "7/7 [==============================] - 0s 13ms/step - loss: 1.2081 - val_loss: 0.8502\n",
      "Epoch 3/100\n",
      "7/7 [==============================] - 0s 12ms/step - loss: 1.0399 - val_loss: 0.6645\n",
      "Epoch 4/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.9018 - val_loss: 0.6900\n",
      "Epoch 5/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.8671 - val_loss: 0.6185\n",
      "Epoch 6/100\n",
      "7/7 [==============================] - 0s 12ms/step - loss: 0.8153 - val_loss: 0.6295\n",
      "Epoch 7/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.7753 - val_loss: 0.5879\n",
      "Epoch 8/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.7580 - val_loss: 0.6194\n",
      "Epoch 9/100\n",
      "7/7 [==============================] - 0s 12ms/step - loss: 0.7473 - val_loss: 0.5877\n",
      "Epoch 10/100\n",
      "7/7 [==============================] - 0s 11ms/step - loss: 0.7016 - val_loss: 0.5615\n",
      "Epoch 11/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6990 - val_loss: 0.5744\n",
      "Epoch 12/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6776 - val_loss: 0.5898\n",
      "Epoch 13/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6909 - val_loss: 0.5614\n",
      "Epoch 14/100\n",
      "7/7 [==============================] - 0s 11ms/step - loss: 0.6706 - val_loss: 0.5547\n",
      "Epoch 15/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6600 - val_loss: 0.5582\n",
      "Epoch 16/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6455 - val_loss: 0.5582\n",
      "Epoch 17/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6286 - val_loss: 0.5475\n",
      "Epoch 18/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6381 - val_loss: 0.5497\n",
      "Epoch 19/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6254 - val_loss: 0.5476\n",
      "Epoch 20/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6018 - val_loss: 0.5461\n",
      "Epoch 21/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.6149 - val_loss: 0.5596\n",
      "Epoch 22/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.6164 - val_loss: 0.5430\n",
      "Epoch 23/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6034 - val_loss: 0.5484\n",
      "Epoch 24/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5994 - val_loss: 0.5493\n",
      "Epoch 25/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5828 - val_loss: 0.5326\n",
      "Epoch 26/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5933 - val_loss: 0.5478\n",
      "Epoch 27/100\n",
      "7/7 [==============================] - 0s 13ms/step - loss: 0.5775 - val_loss: 0.5396\n",
      "Epoch 28/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5868 - val_loss: 0.5393\n",
      "Epoch 29/100\n",
      "7/7 [==============================] - 0s 12ms/step - loss: 0.5833 - val_loss: 0.5451\n",
      "Epoch 30/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5745 - val_loss: 0.5392\n",
      "Epoch 31/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5696 - val_loss: 0.5495\n",
      "Epoch 32/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5751 - val_loss: 0.5345\n",
      "Epoch 33/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5683 - val_loss: 0.5381\n",
      "Epoch 34/100\n",
      "7/7 [==============================] - 0s 12ms/step - loss: 0.5710 - val_loss: 0.5486\n",
      "Epoch 35/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5817 - val_loss: 0.5358\n",
      "\n",
      "********* Model 6 *********\n",
      "Epoch 1/100\n",
      "7/7 [==============================] - 1s 30ms/step - loss: 2.0338 - val_loss: 0.9148\n",
      "Epoch 2/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 1.2817 - val_loss: 0.8721\n",
      "Epoch 3/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.9733 - val_loss: 0.8622\n",
      "Epoch 4/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.9080 - val_loss: 0.8075\n",
      "Epoch 5/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.8260 - val_loss: 0.7380\n",
      "Epoch 6/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.7602 - val_loss: 0.6740\n",
      "Epoch 7/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.7389 - val_loss: 0.7064\n",
      "Epoch 8/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.7059 - val_loss: 0.6726\n",
      "Epoch 9/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6989 - val_loss: 0.7171\n",
      "Epoch 10/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6897 - val_loss: 0.6599\n",
      "Epoch 11/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6859 - val_loss: 0.6818\n",
      "Epoch 12/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.6532 - val_loss: 0.6561\n",
      "Epoch 13/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6469 - val_loss: 0.6416\n",
      "Epoch 14/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6517 - val_loss: 0.6434\n",
      "Epoch 15/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6318 - val_loss: 0.6297\n",
      "Epoch 16/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6152 - val_loss: 0.6353\n",
      "Epoch 17/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6102 - val_loss: 0.6279\n",
      "Epoch 18/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6123 - val_loss: 0.6488\n",
      "Epoch 19/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6246 - val_loss: 0.6427\n",
      "Epoch 20/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5927 - val_loss: 0.6144\n",
      "Epoch 21/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5947 - val_loss: 0.6268\n",
      "Epoch 22/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6011 - val_loss: 0.6210\n",
      "Epoch 23/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5940 - val_loss: 0.6262\n",
      "Epoch 24/100\n",
      "7/7 [==============================] - 0s 12ms/step - loss: 0.5908 - val_loss: 0.6239\n",
      "Epoch 25/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5846 - val_loss: 0.6052\n",
      "Epoch 26/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5798 - val_loss: 0.6354\n",
      "Epoch 27/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5763 - val_loss: 0.6037\n",
      "Epoch 28/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5636 - val_loss: 0.6250\n",
      "Epoch 29/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5619 - val_loss: 0.6168\n",
      "Epoch 30/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5699 - val_loss: 0.6045\n",
      "Epoch 31/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5622 - val_loss: 0.6337\n",
      "Epoch 32/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5687 - val_loss: 0.6070\n",
      "Epoch 33/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5717 - val_loss: 0.6203\n",
      "Epoch 34/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5630 - val_loss: 0.6082\n",
      "Epoch 35/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5599 - val_loss: 0.6223\n",
      "Epoch 36/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5698 - val_loss: 0.6090\n",
      "Epoch 37/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5648 - val_loss: 0.6138\n",
      "\n",
      "********* Model 7 *********\n",
      "Epoch 1/100\n",
      "7/7 [==============================] - 1s 30ms/step - loss: 1.9188 - val_loss: 0.9402\n",
      "Epoch 2/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 1.1985 - val_loss: 0.9136\n",
      "Epoch 3/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 1.0234 - val_loss: 0.8204\n",
      "Epoch 4/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.8973 - val_loss: 0.7640\n",
      "Epoch 5/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.7806 - val_loss: 0.6904\n",
      "Epoch 6/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.7593 - val_loss: 0.7084\n",
      "Epoch 7/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.7539 - val_loss: 0.7037\n",
      "Epoch 8/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.7429 - val_loss: 0.6669\n",
      "Epoch 9/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.7272 - val_loss: 0.6647\n",
      "Epoch 10/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.7010 - val_loss: 0.6523\n",
      "Epoch 11/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.6951 - val_loss: 0.6315\n",
      "Epoch 12/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6705 - val_loss: 0.6216\n",
      "Epoch 13/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.6527 - val_loss: 0.6430\n",
      "Epoch 14/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6480 - val_loss: 0.6027\n",
      "Epoch 15/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6340 - val_loss: 0.6141\n",
      "Epoch 16/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.6122 - val_loss: 0.6011\n",
      "Epoch 17/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.6192 - val_loss: 0.5652\n",
      "Epoch 18/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5946 - val_loss: 0.6162\n",
      "Epoch 19/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.6079 - val_loss: 0.5666\n",
      "Epoch 20/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5823 - val_loss: 0.5730\n",
      "Epoch 21/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5953 - val_loss: 0.5862\n",
      "Epoch 22/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5727 - val_loss: 0.5556\n",
      "Epoch 23/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5797 - val_loss: 0.5673\n",
      "Epoch 24/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5791 - val_loss: 0.5570\n",
      "Epoch 25/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5662 - val_loss: 0.5729\n",
      "Epoch 26/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5734 - val_loss: 0.5620\n",
      "Epoch 27/100\n",
      "7/7 [==============================] - 0s 12ms/step - loss: 0.5596 - val_loss: 0.5646\n",
      "Epoch 28/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5723 - val_loss: 0.5623\n",
      "Epoch 29/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5763 - val_loss: 0.5748\n",
      "Epoch 30/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5630 - val_loss: 0.5729\n",
      "Epoch 31/100\n",
      "7/7 [==============================] - 0s 12ms/step - loss: 0.5734 - val_loss: 0.5632\n",
      "Epoch 32/100\n",
      "7/7 [==============================] - 0s 15ms/step - loss: 0.5742 - val_loss: 0.5652\n",
      "\n",
      "********* Model 8 *********\n",
      "Epoch 1/100\n",
      "7/7 [==============================] - 1s 30ms/step - loss: 1.6196 - val_loss: 1.2195\n",
      "Epoch 2/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 1.1247 - val_loss: 0.9191\n",
      "Epoch 3/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.9828 - val_loss: 0.8123\n",
      "Epoch 4/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.9008 - val_loss: 0.7885\n",
      "Epoch 5/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.8696 - val_loss: 0.7222\n",
      "Epoch 6/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.7899 - val_loss: 0.6708\n",
      "Epoch 7/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.7377 - val_loss: 0.6870\n",
      "Epoch 8/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.7409 - val_loss: 0.6501\n",
      "Epoch 9/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.7060 - val_loss: 0.5986\n",
      "Epoch 10/100\n",
      "7/7 [==============================] - 0s 12ms/step - loss: 0.6798 - val_loss: 0.6123\n",
      "Epoch 11/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6772 - val_loss: 0.6112\n",
      "Epoch 12/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6579 - val_loss: 0.5867\n",
      "Epoch 13/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6476 - val_loss: 0.5749\n",
      "Epoch 14/100\n",
      "7/7 [==============================] - 0s 13ms/step - loss: 0.6454 - val_loss: 0.5774\n",
      "Epoch 15/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6181 - val_loss: 0.5802\n",
      "Epoch 16/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6072 - val_loss: 0.5636\n",
      "Epoch 17/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6159 - val_loss: 0.5621\n",
      "Epoch 18/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6158 - val_loss: 0.5755\n",
      "Epoch 19/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5996 - val_loss: 0.5611\n",
      "Epoch 20/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5927 - val_loss: 0.5768\n",
      "Epoch 21/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5910 - val_loss: 0.5335\n",
      "Epoch 22/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5843 - val_loss: 0.5562\n",
      "Epoch 23/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5769 - val_loss: 0.5514\n",
      "Epoch 24/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5767 - val_loss: 0.5559\n",
      "Epoch 25/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5813 - val_loss: 0.5372\n",
      "Epoch 26/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5698 - val_loss: 0.5937\n",
      "Epoch 27/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5832 - val_loss: 0.5283\n",
      "Epoch 28/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5776 - val_loss: 0.5736\n",
      "Epoch 29/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5803 - val_loss: 0.5513\n",
      "Epoch 30/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5804 - val_loss: 0.5519\n",
      "Epoch 31/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5708 - val_loss: 0.5608\n",
      "Epoch 32/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5722 - val_loss: 0.5539\n",
      "Epoch 33/100\n",
      "7/7 [==============================] - 0s 12ms/step - loss: 0.5716 - val_loss: 0.5572\n",
      "Epoch 34/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5658 - val_loss: 0.5404\n",
      "Epoch 35/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5692 - val_loss: 0.5423\n",
      "Epoch 36/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5661 - val_loss: 0.5346\n",
      "Epoch 37/100\n",
      "7/7 [==============================] - 0s 12ms/step - loss: 0.5643 - val_loss: 0.5500\n",
      "\n",
      "********* Model 9 *********\n",
      "Epoch 1/100\n",
      "7/7 [==============================] - 1s 29ms/step - loss: 2.3050 - val_loss: 1.1329\n",
      "Epoch 2/100\n",
      "7/7 [==============================] - 0s 12ms/step - loss: 1.3831 - val_loss: 0.8234\n",
      "Epoch 3/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 1.0270 - val_loss: 0.8104\n",
      "Epoch 4/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.8832 - val_loss: 0.7813\n",
      "Epoch 5/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.8181 - val_loss: 0.7436\n",
      "Epoch 6/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.7682 - val_loss: 0.7346\n",
      "Epoch 7/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.7434 - val_loss: 0.7015\n",
      "Epoch 8/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.7344 - val_loss: 0.7282\n",
      "Epoch 9/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.7275 - val_loss: 0.6796\n",
      "Epoch 10/100\n",
      "7/7 [==============================] - 0s 12ms/step - loss: 0.7202 - val_loss: 0.6890\n",
      "Epoch 11/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.6842 - val_loss: 0.6863\n",
      "Epoch 12/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.6838 - val_loss: 0.6681\n",
      "Epoch 13/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.6648 - val_loss: 0.6790\n",
      "Epoch 14/100\n",
      "7/7 [==============================] - 0s 12ms/step - loss: 0.6532 - val_loss: 0.6426\n",
      "Epoch 15/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6390 - val_loss: 0.6781\n",
      "Epoch 16/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.6498 - val_loss: 0.6302\n",
      "Epoch 17/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.6229 - val_loss: 0.6379\n",
      "Epoch 18/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6066 - val_loss: 0.6252\n",
      "Epoch 19/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.6003 - val_loss: 0.6216\n",
      "Epoch 20/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5968 - val_loss: 0.6355\n",
      "Epoch 21/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5986 - val_loss: 0.6072\n",
      "Epoch 22/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5929 - val_loss: 0.6223\n",
      "Epoch 23/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5968 - val_loss: 0.6044\n",
      "Epoch 24/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5893 - val_loss: 0.6187\n",
      "Epoch 25/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5872 - val_loss: 0.6147\n",
      "Epoch 26/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5794 - val_loss: 0.6037\n",
      "Epoch 27/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5795 - val_loss: 0.6115\n",
      "Epoch 28/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5868 - val_loss: 0.5975\n",
      "Epoch 29/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5834 - val_loss: 0.6120\n",
      "Epoch 30/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5680 - val_loss: 0.5999\n",
      "Epoch 31/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5631 - val_loss: 0.6100\n",
      "Epoch 32/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5729 - val_loss: 0.6031\n",
      "Epoch 33/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5704 - val_loss: 0.6186\n",
      "Epoch 34/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5741 - val_loss: 0.5918\n",
      "Epoch 35/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5633 - val_loss: 0.6129\n",
      "Epoch 36/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5651 - val_loss: 0.5997\n",
      "Epoch 37/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5543 - val_loss: 0.6167\n",
      "Epoch 38/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5718 - val_loss: 0.6084\n",
      "Epoch 39/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5542 - val_loss: 0.6207\n",
      "Epoch 40/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5618 - val_loss: 0.6081\n",
      "Epoch 41/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5607 - val_loss: 0.6141\n",
      "Epoch 42/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5594 - val_loss: 0.5947\n",
      "Epoch 43/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5604 - val_loss: 0.6149\n",
      "Epoch 44/100\n",
      "7/7 [==============================] - 0s 13ms/step - loss: 0.5724 - val_loss: 0.6219\n",
      "\n",
      "********* Model 10 *********\n",
      "Epoch 1/100\n",
      "7/7 [==============================] - 1s 31ms/step - loss: 2.1971 - val_loss: 0.9929\n",
      "Epoch 2/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 1.2420 - val_loss: 0.7574\n",
      "Epoch 3/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 1.0031 - val_loss: 0.6867\n",
      "Epoch 4/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.9347 - val_loss: 0.6309\n",
      "Epoch 5/100\n",
      "7/7 [==============================] - 0s 11ms/step - loss: 0.8864 - val_loss: 0.5925\n",
      "Epoch 6/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.8045 - val_loss: 0.5658\n",
      "Epoch 7/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.7922 - val_loss: 0.5760\n",
      "Epoch 8/100\n",
      "7/7 [==============================] - 0s 11ms/step - loss: 0.7620 - val_loss: 0.5551\n",
      "Epoch 9/100\n",
      "7/7 [==============================] - 0s 12ms/step - loss: 0.7467 - val_loss: 0.5684\n",
      "Epoch 10/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.7445 - val_loss: 0.5825\n",
      "Epoch 11/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.7213 - val_loss: 0.5495\n",
      "Epoch 12/100\n",
      "7/7 [==============================] - 0s 11ms/step - loss: 0.7328 - val_loss: 0.5910\n",
      "Epoch 13/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6913 - val_loss: 0.5670\n",
      "Epoch 14/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6883 - val_loss: 0.5664\n",
      "Epoch 15/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6965 - val_loss: 0.5410\n",
      "Epoch 16/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6665 - val_loss: 0.5447\n",
      "Epoch 17/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6623 - val_loss: 0.5331\n",
      "Epoch 18/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6400 - val_loss: 0.5231\n",
      "Epoch 19/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6327 - val_loss: 0.5262\n",
      "Epoch 20/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6260 - val_loss: 0.5097\n",
      "Epoch 21/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6080 - val_loss: 0.5309\n",
      "Epoch 22/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6026 - val_loss: 0.5281\n",
      "Epoch 23/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6037 - val_loss: 0.5175\n",
      "Epoch 24/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.6051 - val_loss: 0.5200\n",
      "Epoch 25/100\n",
      "7/7 [==============================] - 0s 13ms/step - loss: 0.6008 - val_loss: 0.5152\n",
      "Epoch 26/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.6042 - val_loss: 0.5223\n",
      "Epoch 27/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5935 - val_loss: 0.5297\n",
      "Epoch 28/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5973 - val_loss: 0.5058\n",
      "Epoch 29/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5812 - val_loss: 0.5284\n",
      "Epoch 30/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5900 - val_loss: 0.5111\n",
      "Epoch 31/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5835 - val_loss: 0.5239\n",
      "Epoch 32/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5828 - val_loss: 0.5149\n",
      "Epoch 33/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5801 - val_loss: 0.5181\n",
      "Epoch 34/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5828 - val_loss: 0.5188\n",
      "Epoch 35/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5764 - val_loss: 0.5082\n",
      "Epoch 36/100\n",
      "7/7 [==============================] - 0s 9ms/step - loss: 0.5678 - val_loss: 0.5301\n",
      "Epoch 37/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5738 - val_loss: 0.5048\n",
      "Epoch 38/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5813 - val_loss: 0.5199\n",
      "Epoch 39/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5538 - val_loss: 0.4991\n",
      "Epoch 40/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5640 - val_loss: 0.5116\n",
      "Epoch 41/100\n",
      "7/7 [==============================] - 0s 13ms/step - loss: 0.5713 - val_loss: 0.5098\n",
      "Epoch 42/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5659 - val_loss: 0.5229\n",
      "Epoch 43/100\n",
      "7/7 [==============================] - 0s 13ms/step - loss: 0.5716 - val_loss: 0.5186\n",
      "Epoch 44/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5576 - val_loss: 0.5070\n",
      "Epoch 45/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5640 - val_loss: 0.5130\n",
      "Epoch 46/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5651 - val_loss: 0.5138\n",
      "Epoch 47/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5752 - val_loss: 0.5106\n",
      "Epoch 48/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5625 - val_loss: 0.5270\n",
      "Epoch 49/100\n",
      "7/7 [==============================] - 0s 10ms/step - loss: 0.5620 - val_loss: 0.5177\n",
      "\n"
     ]
    }
   ],
   "source": [
    "keras.backend.clear_session()\n",
    "n_ensemble = 10\n",
    "models = [None]*n_ensemble\n",
    "\n",
    "for i in range(n_ensemble):\n",
    "    print(f'********* Model {i+1} *********')\n",
    "    models[i] = build_and_fit(best,X,y,verbose=True,resample_seed=i)\n",
    "    print()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "77f03e7f",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-03-07T16:03:29.494032Z",
     "iopub.status.busy": "2023-03-07T16:03:29.492468Z",
     "iopub.status.idle": "2023-03-07T16:03:29.814762Z",
     "shell.execute_reply": "2023-03-07T16:03:29.813217Z"
    },
    "papermill": {
     "duration": 0.421558,
     "end_time": "2023-03-07T16:03:29.817618",
     "exception": false,
     "start_time": "2023-03-07T16:03:29.396060",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model\"\n",
      "__________________________________________________________________________________________________\n",
      " Layer (type)                   Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      " input_1 (InputLayer)           [(None, 8)]          0           []                               \n",
      "                                                                                                  \n",
      " sequential (Sequential)        (None, 1)            6559        ['input_1[0][0]']                \n",
      "                                                                                                  \n",
      " sequential_1 (Sequential)      (None, 1)            6559        ['input_1[0][0]']                \n",
      "                                                                                                  \n",
      " sequential_2 (Sequential)      (None, 1)            6559        ['input_1[0][0]']                \n",
      "                                                                                                  \n",
      " sequential_3 (Sequential)      (None, 1)            6559        ['input_1[0][0]']                \n",
      "                                                                                                  \n",
      " sequential_4 (Sequential)      (None, 1)            6559        ['input_1[0][0]']                \n",
      "                                                                                                  \n",
      " sequential_5 (Sequential)      (None, 1)            6559        ['input_1[0][0]']                \n",
      "                                                                                                  \n",
      " sequential_6 (Sequential)      (None, 1)            6559        ['input_1[0][0]']                \n",
      "                                                                                                  \n",
      " sequential_7 (Sequential)      (None, 1)            6559        ['input_1[0][0]']                \n",
      "                                                                                                  \n",
      " sequential_8 (Sequential)      (None, 1)            6559        ['input_1[0][0]']                \n",
      "                                                                                                  \n",
      " sequential_9 (Sequential)      (None, 1)            6559        ['input_1[0][0]']                \n",
      "                                                                                                  \n",
      " average (Average)              (None, 1)            0           ['sequential[0][0]',             \n",
      "                                                                  'sequential_1[0][0]',           \n",
      "                                                                  'sequential_2[0][0]',           \n",
      "                                                                  'sequential_3[0][0]',           \n",
      "                                                                  'sequential_4[0][0]',           \n",
      "                                                                  'sequential_5[0][0]',           \n",
      "                                                                  'sequential_6[0][0]',           \n",
      "                                                                  'sequential_7[0][0]',           \n",
      "                                                                  'sequential_8[0][0]',           \n",
      "                                                                  'sequential_9[0][0]']           \n",
      "                                                                                                  \n",
      "==================================================================================================\n",
      "Total params: 65,590\n",
      "Trainable params: 65,590\n",
      "Non-trainable params: 0\n",
      "__________________________________________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "# Create ensemble model in Keras\n",
    "model_input = tf.keras.Input(shape=(X.shape[1],))\n",
    "model_outputs = [model(model_input) for model in models]\n",
    "ensemble_output = tf.keras.layers.Average()(model_outputs)\n",
    "ensemble_model = tf.keras.Model(inputs=model_input, outputs=ensemble_output)\n",
    "\n",
    "print(ensemble_model.summary())\n",
    "\n",
    "ensemble_model.save('ensemble_model.h5')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "33970582",
   "metadata": {
    "papermill": {
     "duration": 0.094948,
     "end_time": "2023-03-07T16:03:30.010402",
     "exception": false,
     "start_time": "2023-03-07T16:03:29.915454",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Generate test predictions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "00c47aae",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-03-07T16:03:30.201115Z",
     "iopub.status.busy": "2023-03-07T16:03:30.200146Z",
     "iopub.status.idle": "2023-03-07T16:03:31.225642Z",
     "shell.execute_reply": "2023-03-07T16:03:31.224161Z"
    },
    "papermill": {
     "duration": 1.125668,
     "end_time": "2023-03-07T16:03:31.228887",
     "exception": false,
     "start_time": "2023-03-07T16:03:30.103219",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "113/113 [==============================] - 1s 2ms/step\n"
     ]
    }
   ],
   "source": [
    "X_test = input_sc.transform(test.drop('id',axis=1).values)\n",
    "y_test_pred = ensemble_model.predict(X_test)\n",
    "\n",
    "y_test_pred_orig = output_sc.inverse_transform(y_test_pred.reshape(-1,1)).ravel()\n",
    "\n",
    "submission = pd.DataFrame({'id':test['id'].values,'Strength':y_test_pred_orig})\n",
    "submission.to_csv('submission.csv',index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "61201609",
   "metadata": {
    "papermill": {
     "duration": 0.094098,
     "end_time": "2023-03-07T16:03:31.502869",
     "exception": false,
     "start_time": "2023-03-07T16:03:31.408771",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.7.12"
  },
  "papermill": {
   "default_parameters": {},
   "duration": 1400.218422,
   "end_time": "2023-03-07T16:03:34.225378",
   "environment_variables": {},
   "exception": null,
   "input_path": "__notebook__.ipynb",
   "output_path": "__notebook__.ipynb",
   "parameters": {},
   "start_time": "2023-03-07T15:40:14.006956",
   "version": "2.4.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}