model_neural_networks.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Neural Network Model\n",
    "This notebook aims to build an artificial neural network model with parameters selected by GridSearchCV. This file was executed on AWS."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "# Import packages\n",
    "from keras.callbacks import ModelCheckpoint\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dropout, Dense, Activation, Flatten\n",
    "from keras.wrappers.scikit_learn import KerasClassifier\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.ensemble import RandomForestRegressor\n",
    "from sklearn.metrics import mean_absolute_error, mean_squared_error, make_scorer\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "import seaborn as sb\n",
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import warnings \n",
    "import math\n",
    "warnings.filterwarnings('ignore')\n",
    "warnings.filterwarnings('ignore', category=DeprecationWarning)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Importing of Files"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Import dataset\n",
    "df = pd.read_csv('data/features_df.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Split the data into training and test sets (30% held out for testing)\n",
    "y = df.loc[:, ['viewCount']] # dependent variable\n",
    "X = df.loc[:, ['titleLen', 'subscriberCount', 'avgViewCount', 'humanCount', 'HOW TO & STYLE', 'SPORTS', 'TRAVEL', 'Negative', 'titleINTJ']] # selected independant variables\n",
    "\n",
    "# Feature Scaling\n",
    "scaler = MinMaxScaler()\n",
    "X = pd.DataFrame(scaler.fit_transform(X),columns=['titleLen', 'subscriberCount', 'avgViewCount', 'humanCount', 'HOW TO & STYLE', 'SPORTS', 'TRAVEL', 'Negative', 'titleINTJ'])\n",
    "\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.30, random_state=123)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Performing Grid Search"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def create_model(neurons=32, dropout_rate=0):\n",
    "    # Setting default values\n",
    "    activation='relu'\n",
    "    init_mode='uniform'\n",
    "    optimizer='adam'\n",
    "    \n",
    "    # Create model\n",
    "    model = Sequential()\n",
    "    model.add(Dense(128, input_dim=X_train.shape[1], kernel_initializer=init_mode, activation=activation))\n",
    "    model.add(Dense(neurons, kernel_initializer=init_mode, activation=activation))\n",
    "    model.add(Dense(neurons, kernel_initializer=init_mode, activation=activation))\n",
    "    model.add(Dense(neurons, kernel_initializer=init_mode, activation=activation))\n",
    "    model.add(Dense(1, kernel_initializer=init_mode, activation='linear'))\n",
    "    \n",
    "    # Compile model\n",
    "    model.compile(loss='mse', optimizer=optimizer, metrics=['mae', 'mse'])\n",
    "    \n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create model\n",
    "model = KerasClassifier(build_fn=create_model, batch_size=1000, epochs=1, verbose=0) #set verbose=1 to see training progress"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Setting parameters for Grid Search\n",
    "parameters = {'epochs':[100],\n",
    "              'batch_size':[1000, 2500],\n",
    "              'neurons':[32, 64, 128]\n",
    "             }"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Run Grid Search\n",
    "mse = make_scorer(mean_squared_error, greater_is_better=False)\n",
    "grid = GridSearchCV(estimator=model, param_grid=parameters, n_jobs=-1, scoring=mse)\n",
    "grid_result = grid.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best: -489874718973712.375000 using {'batch_size': 1000, 'epochs': 100, 'neurons': 32}\n"
     ]
    }
   ],
   "source": [
    "# Print Best Results\n",
    "print(\"Best: %f using %s\" % (grid_result.best_score_, grid_result.best_params_))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Print All Results\n",
    "# means = grid_result.cv_results_['mean_test_score']\n",
    "# stds = grid_result.cv_results_['std_test_score']\n",
    "# params = grid_result.cv_results_['params']\n",
    "# for mean, stdev, param in zip(means, stds, params):\n",
    "#     print(\"%f (%f) with: %r\" % (mean, stdev, param))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Building the Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_1\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense_1 (Dense)              (None, 128)               1280      \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 32)                4128      \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 32)                1056      \n",
      "_________________________________________________________________\n",
      "dense_4 (Dense)              (None, 32)                1056      \n",
      "_________________________________________________________________\n",
      "dense_5 (Dense)              (None, 1)                 33        \n",
      "=================================================================\n",
      "Total params: 7,553\n",
      "Trainable params: 7,553\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "nnModel = Sequential()\n",
    "\n",
    "# The Input Layer :\n",
    "nnModel.add(Dense(128, kernel_initializer='normal',input_dim = X_train.shape[1], activation='relu'))\n",
    "\n",
    "# The Hidden Layers :\n",
    "nnModel.add(Dense(32, kernel_initializer='normal',activation='relu'))\n",
    "nnModel.add(Dense(32, kernel_initializer='normal',activation='relu'))\n",
    "nnModel.add(Dense(32, kernel_initializer='normal',activation='relu'))\n",
    "\n",
    "# The Output Layer :\n",
    "nnModel.add(Dense(1, kernel_initializer='normal',activation='linear'))\n",
    "\n",
    "# Compile the network :\n",
    "nnModel.compile(loss='mse', optimizer='adam', metrics =['mae', 'mse'])\n",
    "nnModel.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "checkpoint_name = 'neural_networks_weights.hdf5' \n",
    "checkpoint = ModelCheckpoint(checkpoint_name, monitor='val_loss', verbose = 0, save_best_only = True, mode ='auto')\n",
    "callbacks_list = [checkpoint]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "history = nnModel.fit(X_train, y_train, epochs=100, batch_size=1000, validation_data = (X_test,y_test), callbacks=callbacks_list, verbose=0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Checking Epoch Progress"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXgAAAEWCAYAAABsY4yMAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjMsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+AADFEAAAgAElEQVR4nO3de5gcdZ3v8fe3e+6TmUwyCQkkhFwEJEQTkgEJICggAq7iriAgLJBFc+R4BFZZD/ucs4+ge87CPt64KQYNgrqgclk5KAgiwipymWCEkBACIZCBxJncL5NkZrq/54+qnqnpzDUzNT1T83k9T2e6q6rr962uzqd+XV1dZe6OiIgkT6rQBYiISDwU8CIiCaWAFxFJKAW8iEhCKeBFRBJKAS8iklAKeBm1zGy6mbmZFfVh2svM7A9DUZfIYFHAy4hgZuvMrMXMJuQNXx6G9PTCVNZpQ/Fi3vAJYc3rIsNOMrNnzGy7mW0xsz+a2bHhuMvMLGNmu/JuhwzxIklCKOBlJHkTuDD3wMzeB5QXrpz9VJrZnMjjzxDUDICZVQMPA7cA44EpwPXAvshz/uTuY/Ju7w5B7ZJACngZSX4MXBJ5fClwd3QCMxtrZnebWZOZvWVm/9vMUuG4tJl9w8w2mdla4GNdPPeHZrbBzN4xs381s3Q/67s08viSvPqOAHD3e9w94+573P0xd3+pH22I9JkCXkaSZ4FqMzsqDN7zgZ/kTXMLMBaYCZxCELKLwnGfA/4GOAaoA87Ne+5dQBvwnnCaM4DP9qO+nwAXhBuSo4Aq4LnI+NeAjJndZWZnmdm4fsxbpN+GXcCb2VIzazSzFX2Y9mQze9HM2sws/z8rZlYd9sRujadaKYBcL/4jwKvAO7kRkdD/Z3ff6e7rgG8Cfx9O8mngO+6+3t23AP8Wee4k4Czganff7e6NwLeBC/pRWwOwGjidLj5duPsO4CTAgTuAJjN7KGw753gz2xa5vdGP9kU6GXYBD/wIOLOP074NXAb8Rzfjvw48NfCSZBj5McG+7cvIC1BgAlACvBUZ9hbBvm6AQ4D1eeNyDgOKgQ25cAW+DxzUz/ruDmu7kP0/XeDuq9z9MnefCswJa/pOZJJn3b0mcpvVz/ZF2g27gHf3p4Et0WFmNsvMHjWzZWb2X2b23nDadeH+y2z+fMxsATAJeGwo6pah4e5vEXxxeTbwQN7oTUArQVjnTKOjl78BODRvXM56gi87J0TCtdrdj+5nifcT7NtfG9ba07K8StChmdPTdCIHatgFfDeWAF909wXANcB3e5o4/FLtm8A/DUFtMvQuB051993Rge6eAX4O/B8zqzKzw4Av0dGT/jlwpZlNDfd/Xxt57gaCzsA3w117qbBjcUp/CgtrOpUu9t2b2XvN7MtmNjV8fChBT//Z/rQh0lfDPuDNbAxwAvALM1tO8LH54F6e9t+BX7v7+l6mkxHI3d9w9/puRn8R2A2sBf5AsPtuaTjuDuA3wF+AF9n/E8AlBLt4VgJbgfvo/b3WVX317t7VvvOdwAeA58xsN0GwrwC+HJlmYRfHwR/b3xpEAGw4XvAj/NHKw+4+Jzx2eLW7d/sfzcx+FE5/X/j4p8AHCXbdjCH4T/tdd7+2u3mIiCTNsO/Bh0cevGlm5wFYYG4vz7nI3ae5+3SCXTp3K9xFZLQZdgFvZvcAfwKONLMGM7scuAi43Mz+ArwCnBNOe6yZNQDnAd83s1cKVbeIyHAzLHfRiIjIwA27HryIiAyOXk+TOpQmTJjg06dPL3QZIiIjxrJlyza5+8Suxg2rgJ8+fTr19d0d/SYiIvnMrNsf1GkXjYhIQingRUQSSgEvIpJQw2offFdaW1tpaGhg7969hS5lSJSVlTF16lSKi4sLXYqIjHDDPuAbGhqoqqpi+vTpmFmhy4mVu7N582YaGhqYMWNGocsRkRFu2O+i2bt3L7W1tYkPdwAzo7a2dtR8WhGReA37gAdGRbjnjKZlFZF4jYiA79XOjdDSXOgqRESGlZEf8Jk22L0JNr0W/B3Ec+ts3ryZefPmMW/ePCZPnsyUKVPaH7e0tPRpHosWLWL16tWDVpOISF8N+y9Ze5UugolHwta3YPt6aNkNY6dCKj3gWdfW1rJ8+XIArrvuOsaMGcM111zTaRp3x91JpbreVt55550DrkNE5ECM/B48QLoYamfBmMmwZwtsfBk2vw67GqFt36A39/rrrzNnzhw+//nPM3/+fDZs2MDixYupq6vj6KOP5mtf+1r7tCeddBLLly+nra2Nmpoarr32WubOncvChQtpbGwc9NpERHJGVA/++v/3Civf3dHzRJ6BbBtkN4OH1+K2VNCjTxWBde7Zzz6kmq9+vL/XVYaVK1dy5513cvvttwNwww03MH78eNra2vjwhz/Mueeey+zZszs9Z/v27ZxyyinccMMNfOlLX2Lp0qVce62uQyIi8UhGDz7K0pAuheKK4JYuBTPItELrnuCWbRtwM7NmzeLYYzsulXnPPfcwf/585s+fz6pVq1i5cuV+zykvL+ess84CYMGCBaxbt27AdYiIdGdE9eAPpKfdLpuB5s2wuwkyLUH4j5sBRSUHNLvKysr2+2vWrOGmm27i+eefp6amhosvvrjLY9lLSjraSqfTtLUNfEMjItKd5PXgu5NKw5iD4KCjoGYatO2FTauDL2UHaMeOHVRVVVFdXc2GDRv4zW9+MwgFi4gMzIjqwQ8KS0FFbdCD37IWNq0JAr9i/AHPcv78+cyePZs5c+Ywc+ZMTjzxxEEsWETkwAyra7LW1dV5/gU/Vq1axVFHHRVPg5k22PomtOyCcdOhfFw87fRTrMssIoliZsvcva6rcaNnF01X0kUwfhYUVwbH0e/bWeiKREQGzegOeIBUCsbPDL5s3fJmcJSNiEgCKOChoydvKdj8Riw/jhIRGWoK+JyiUqidGfw4avPr0Na3c82IiAxXCvio4orglAfZTBDymdZCVyQicsAU8PlKKoN98tnWMOTVkxeRkUkB35XSMTB+JpubGpk39/3Mmzv3gE8XDLB06VI2btwYY8EiIvsbfT906qvSKmoPr2P5b38BwHW33cuYseP2O11wXyxdupT58+czefLkwa5SRKRbCvielFRC7eHBrprmLVBR1j7qrrvu4rbbbqOlpYUTTjiBW2+9lWw2y6JFi1i+fDnuzuLFi5k0aRLLly/n/PPPp7y8nOeff77TOWlEROISa8Cb2VXA5wAD7nD37wxoho9cG5zrfTBNfh+cdUP344vLYMIRwbls9myB7etZ8dZWHnzwQZ555hmKiopYvHgx9957L7NmzWLTpk28/HJQ47Zt26ipqeGWW27h1ltvZd68eYNbu4hID2ILeDObQxDuxwEtwKNm9it3XxNXm7EpKgnPX5OF3Zv47S9/zgsvPE9dXfDr4D179nDooYfy0Y9+lNWrV3PVVVdx9tlnc8YZZxS4cBEZzeLswR8FPOvuzQBm9hTwt8C/H/Ace+ppx80MysbC+Jl4NsM/nPcxvn7dv0DVwcEVpUIvvfQSjzzyCDfffDP3338/S5YsKVzNIjKqxXkUzQrgZDOrNbMK4Gzg0PyJzGyxmdWbWX1TU1OM5QySsrGc/skL+fmvfsemhjegcRWb163k7XVv0tTUhLtz3nnncf311/Piiy8CUFVVxc6dOs+NiAyt2Hrw7r7KzG4EHgd2AX8B9rvChbsvAZZAcDbJuOoZTO+bewxfvf7rnH7R1WTb9lGcNm6/8V9Ij5nA5Vd+BXcwM2688UYAFi1axGc/+1l9ySoiQ2rIThdsZv8XaHD373Y3zZCfLniwtDTDzg2wbwdgUFYDlbVQMibYtdNPI2KZRWRY6Ol0wXEfRXOQuzea2TTg74CFcbZXMCXhKQ5a9wSXBWzeAnu3BteHLauG0uog7NPFBxT4IiIHIu7j4O83s1qgFfiCu2+Nub3CKi6HsVOh6hDYtx327gh69XvCxU4VBxuD4vLgYuBFZZAuCQ7BVPCLyCCLNeDd/YODNB9sJAVgKhVcHap8HLgHPfvW3cH1X1uaYe/2vCdY0LtPFeFWFHwKeOLrwWUEy8dDeQ2UVgWfAsqqg11ApdXBaY5FRLox7BOirKyMzZs3U1tbO7JCPscs6LWXVEDlxGBYNguZfcF55zMtwYnNMq14Wyubt++kbOtq+OO3glMX96S4Mvg0UBzOv7QqCP7SqmB4UVn4N/JpoagsOK4/Xdq+USGVDoeXQlF5MMxSQe2povB5JeHwdPiJI/c3Fc6jqPPzossvIgUx7AN+6tSpNDQ0MCIOoRwEZWVjmXrKJXDqomA3T/MW2LsN9u0Krh3bvttnW/C3tTn4hNCyK7jk4N5tsH09tO6FtsitkGfFtFRwwyIbAOsh/K1jmk7zsGCjQjcbEHcg/6AB23+66PzNIrXlhkVn48GG1rPB/C3VsWEjMs1+d72j3dzGD4JTUXum9/r2E60jMlnutc0tA97N6xCdVTaoI5vpWP6ebrnneJZOjeevo/bliLyA3R7EEdaZm2/76xrurozOx8lrl45hXc2/q/dV9DXJ3c91oNIlHZ2dTCtk24Ibkdcm+vpjkd2qkfXXPt/IMkZfq9x7uH0+hMPSwaf1i+/v5rU6cMM+4IuLi5kxY0ahyyiM3G6ewZDNhkG/L7iYSWZf8CbOtIWfIFrCjcKeYFrPBkGUzQTjMi3hGz8T/M2FhGciw/PCKxeKeDAu+p86+p9tv5Am7z9u7nlhG9Hh0eflbzja55E3XXvbkfrcO9+P/udNRTYAng1fn0zHeOg6oNunbwuCo/0/eToSCF3VF12+yHzbNyyRIM+9ltHwyf8UFeUe+QQWhnduedoDN3o/GkQpOgdZfrjmDWuvoZtaOs0z936KrpPc8kQ3JHnrp7+fEHPTR1/H9vd4axj0xWFHgs4boE4bHY+8F7uqKS/42zfO2c6dlNywsur+LUcfDfuAl0GSSgW7cagodCUiMkR0PngRkYRSwIuIJJQCXkQkoRTwIiIJpYAXEUkoBbyISEIp4EVEEkoBLyKSUAp4EZGEUsCLiCSUAl5EJKEU8CIiCaWAFxFJKAW8iEhCKeBFRBJKAS8iklAKeBGRhFLAi4gklAJeRCShFPAiIgmlgBcRSSgFvIhIQingRUQSKtaAN7N/NLNXzGyFmd1jZmVxticiIh1iC3gzmwJcCdS5+xwgDVwQV3siItJZ3LtoioByMysCKoB3Y25PRERCsQW8u78DfAN4G9gAbHf3x/KnM7PFZlZvZvVNTU1xlSMiMurEuYtmHHAOMAM4BKg0s4vzp3P3Je5e5+51EydOjKscEZFRJ85dNKcDb7p7k7u3Ag8AJ8TYnoiIRMQZ8G8Dx5tZhZkZcBqwKsb2REQkIs598M8B9wEvAi+HbS2Jqz0REemsKM6Zu/tXga/G2YaIiHRNv2QVEUkoBbyISEIp4EVEEkoBLyKSUAp4EZGEUsCLiCSUAl5EJKEU8CIiCaWAFxFJKAW8iEhCKeBFRBJKAS8iklAKeBGRhFLAi4gklAJeRCShFPAiIgmlgBcRSSgFvIhIQingRUQSSgEvIpJQCngRkYRSwIuIJJQCXkQkoRTwIiIJpYAXEUkoBbyISEIp4EVEEkoBLyKSUD0GvJldHLl/Yt64/9HLc480s+WR2w4zu3pg5YqISF/11oP/UuT+LXnj/qGnJ7r7anef5+7zgAVAM/Bg/0sUEZED0VvAWzf3u3rck9OAN9z9rX48R0REBqC3gPdu7nf1uCcXAPd0NcLMFptZvZnVNzU19WOWIiLSE3PvPqfNrBl4naC3Piu8T/h4prtX9tqAWQnwLnC0u/+1p2nr6uq8vr6+j6WLiIiZLXP3uq7GFfXy3KMGof2zgBd7C3cRERlcPQZ8/j5zM6sFTgbedvdlfWzjQrrZPSMiIvHp7TDJh81sTnj/YGAFwdEzP+7LIY9mVgF8BHhgEGoVEZF+6O1L1hnuviK8vwh43N0/DnyAXg6TBHD3ZnevdfftA6xTRET6qbeAb43cPw34NYC77wSycRUlIiID19uXrOvN7ItAAzAfeBTAzMqB4phrExGRAeitB385cDRwGXC+u28Lhx8P3BljXSIiMkC9HUXTCHy+i+FPAk/GVZSIiAxcjwFvZg/1NN7dPzG45YiIyGDpbR/8QmA9wXHsz9G/88+IiEgB9RbwkwmOY78Q+AzwK+Aed38l7sJERGRgevyS1d0z7v6ou19K8MXq68DvwyNrRERkGOutB4+ZlQIfI+jFTwduRr9MFREZ9nr7kvUuYA7wCHB95FetIiIyzPXWg/97YDdwBHClWft3rAa4u1fHWJuIiAxAb8fB66LcIiIjlAJcRCShFPAiIgmlgBcRSSgFvIhIQingRUQSSgEvIpJQCngRkYRSwIuIJJQCXkQkoRTwIiIJpYAXEUkoBbyISEIp4EVEEkoBLyKSUAp4EZGEijXgzazGzO4zs1fNbJWZLYyzPRER6dDrNVkH6CbgUXc/18xKgIqY2xMRkVBsAW9m1cDJwGUA7t4CtMTVnoiIdBbnLpqZQBNwp5n92cx+YGaV+ROZ2WIzqzez+qamphjLEREZXeIM+CJgPvA9dz+G4OLd1+ZP5O5L3L3O3esmTpwYYzkiIqNLnAHfADS4+3Ph4/sIAl9ERIZAbAHv7huB9WZ2ZDjoNGBlXO2JiEhncR9F80Xgp+ERNGuBRTG3JyIioVgD3t2XA3VxtiEiIl3TL1lFRBJKAS8iklAKeBGRhFLAi4gklAJeRCShFPAiIgmlgBcRSSgFvIhIQingRUQSSgEvIpJQCngRkYRSwIuIJJQCXkQkoRTwIiIJpYAXEUkoBbyISEIp4EVEEkoBLyKSUAp4EZGEUsCLiCSUAl5EJKEU8CIiCaWAFxFJKAW8iEhCKeBFRBJKAS8iklAKeBGRhCqKc+Zmtg7YCWSANnevi7M9ERHpEGvAhz7s7puGoB0REYnQLhoRkYSKO+AdeMzMlpnZ4q4mMLPFZlZvZvVNTU0xlyMiMnrEHfAnuvt84CzgC2Z2cv4E7r7E3evcvW7ixIkxlyMiMnrEGvDu/m74txF4EDguzvZERKRDbAFvZpVmVpW7D5wBrIijrWt+8Re+/9QbrN/SHMfsRURGpDiPopkEPGhmuXb+w90fHexGmlvaeO2vO7lvWQP/9sirzJ06lguOm8a5C6ZSnNZ3yCIyepm7F7qGdnV1dV5fX39Az12/pZlfv7yBXy5/l5UbdnBYbQVXn344n5g7hXTKBrlSEZHhwcyWdfcbo8QEfI678+TqRr7xm9dYuWEHNRXFzJ82jgWHjWPhrFrmTa0hpcAXkYToKeCH4odOQ8rMOPW9k/jQEQfx2Mq/8uSrjSx7eyu/e7URgIPHlnHmnMl8av5U5kwZW+BqRUTik7gefHe27m7h96818uuXN/LUa020tGX5xNxD+MqZRzJ1XEUsbYqIxG1U7aLpix17W7nj6bUseXotDlxxyiyuOu1w7boRkRGnp4AflYeZVJcV8+UzjuTJaz7EmUdP5qYn1nD1z5bT0pYtdGkiIoMmcfvg++OQmnJuumAeRx1czY2PvsrW5ha+d/ECxpSO6pdFRBJiVPbgo8yMKz40i38/9/0888ZmPnPHs2xrbil0WSIiAzbqAz7n03WH8v2LF/Dqhp1c9IPn2LpbIS8iI5sCPuL02ZNYcskC1jTu4jM/eI4tCnkRGcEU8Hk+dORB3HFJHWubdvGZO56lcefeQpckInJAFPBdOOWIifzw0mN5a3Mz593+J53ETERGJAV8N046fAI//dwH2L6nlU997xle3bij0CWJiPSLAr4H86eN4xf/bSEpMz59+5/4/erGQpckItJnCvheHD6pivuuWMiUcRUs+tEL3PLEGrLZ4fPrXxGR7ijg+2DquAoeuOIEzpl7CN98/DUW/7iezbv2FbosEZEeKeD7qLwkzbfPn8d1H5/NU6818ZFvP81//vkdhtO5fEREohTw/WBmXHbiDB7+4geZNr6Cq3+2nEU/eoE3N+0udGkiIvtRwB+AIydXcf8VJ/DVj8/m+Te38JFvPcV1D72i3TYiMqyMytMFD6bGnXu56bdruPeF9VQUp7nkhMO47IQZTKwqLXRpIjIK6HzwQ+D1xp186/HXeGTFRorTKc5dMJWLPjCN2QdXE154XERk0Cngh9Dapl3c8V9ruX/ZO7Rkshx+0Bg+ecwUzpg9ifccNEZhLyKDSgFfAFt2t/Crlzfwyz+/Q/1bWwGYVF3KSe+ZyHEzxjFnyliOmFRFcVpfg4jIgVPAF9g72/bw9GtN/GHNJv74xia2NbcCUFKU4j0TxzBjYiUzJ1Ry6PgKDhlbzuSxZUyqLmVMaZF6/CLSIwX8MJLNOm9taealhm2seGc7axp38eam3azf0kz+D2RL0inGV5YwvrKE2jEl1FaWMK6yhLHlxVSXFVNdXszY8mJqKoK/laVFjCkporI0TZE+GYiMCj0FvK5NN8RSKWPGhEpmTKjknHlT2oe3tGXZuH0v727fw4bte2jcsY8tzS1s2dXClt0tbN7dwrrNu9m6u5Vd+9p6backnaKsOEV5SZry4jTlJUWUF6coK05TWpSitChNaXGq/X5JUSq4pYO/xWmjKJWiKG2YGSmD4lSK4iKjJJ2mKG0Up410KkVRykiF06RShgHBB49wmAXjzYLhKTPS4XPS4fTQMS73Ohl0PC+YHUbHY7POz21/ZLn295+uO7m2La8NgKxD1h13Oi1DR9t5dehTlwwTCvhhoqQoxbTaCqbVVvQ6bVsmy659bWzf09rptntfGzv3trF7X4Y9rRn2tmZobmljb2uW5pYMe1rbaGkLnru3NUNLW5Z9bVn2tmZozTgtbVlaMrrweFyiuZ//wTllBBu7cEOZ28i4g+PhX8J/OvPIwNzz02FjuTHRjWXH87qokWDj6k6nX2mbdb9h7fzs/avrm47OgONk29u39tem62ft30quk5DbGHclt2j56yH3nPyOhHtYE94+LhUZng1nFH1urr78DX5HZ6Vj3PiKEn7++YW9vEb9p4AfgYrSKWoqSqipKBn0ebs7rRmnNZOlNZMlk+14A7dlg41AayZLS1uWtqyTyWZpy3RMk+vpQtjrDeeZyQZ/nWA3VdahLZtt/48RBErwHzWYh+c9zgVcbp4dAZR73HHfI8vTOQBzvfB82bD9aNu5+QXB2/EJIevefsK5zm17p8DolB3Rmsj9J+8oJHhNnEy4oJ0/Mdh+n2Dyl8citWTdyWS9UwgHwzuCiMi4/BKz7vsFk0dfd7zT9Psta56+fJ7peJ94uCHqWM5sNlyevBnl3hP5y5l7z3RXVMfrn6svspy5jWmn5SXc6Ha8Jtkw2HMbpeiGpqv3Zf5y5m+sq8riiWIFvHRiZpQUGSVF2ocvMtLF/r/YzNJm9mczezjutkREpMNQdNOuAlYNQTsiIhIRa8Cb2VTgY8AP4mxHRET2F3cP/jvAV4BuD80ws8VmVm9m9U1NTTGXIyIyesQW8Gb2N0Cjuy/raTp3X+Lude5eN3HixLjKEREZdeLswZ8IfMLM1gH3Aqea2U9ibE9ERCJiC3h3/2d3n+ru04ELgN+5+8VxtSciIp3pYGcRkYQaVicbM7Mm4K0DfPoEYNMgljMSjMZlhtG53KNxmWF0Lnd/l/kwd+/yC8xhFfADYWb13Z1RLalG4zLD6Fzu0bjMMDqXezCXWbtoREQSSgEvIpJQSQr4JYUuoABG4zLD6Fzu0bjMMDqXe9CWOTH74EVEpLMk9eBFRCRCAS8iklAjPuDN7EwzW21mr5vZtYWuJy5mdqiZPWlmq8zsFTO7Khw+3sweN7M14d9xha51sOVfU8DMZpjZc+Ey/8zMBv/SVgVmZjVmdp+ZvRqu84VJX9dm9o/he3uFmd1jZmVJXNdmttTMGs1sRWRYl+vWAjeH+faSmc3vT1sjOuDNLA3cBpwFzAYuNLPZha0qNm3Al939KOB44Avhsl4LPOHuhwNPhI+TJv+aAjcC3w6XeStweUGqitdNwKPu/l5gLsHyJ3Zdm9kU4Eqgzt3nAGmCU5wkcV3/CDgzb1h36/Ys4PDwthj4Xn8aGtEBDxwHvO7ua929heCkZucUuKZYuPsGd38xvL+T4D/8FILlvSuc7C7gk4WpMB751xSw4KKYpwL3hZMkcZmrgZOBHwK4e4u7byPh65rgEqLlZlYEVAAbSOC6dvengS15g7tbt+cAd3vgWaDGzA7ua1sjPeCnAOsjjxvCYYlmZtOBY4DngEnuvgGCjQBwUOEqi0X+NQVqgW3u3hY+TuI6nwk0AXeGu6Z+YGaVJHhdu/s7wDeAtwmCfTuwjOSv65zu1u2AMm6kB3xXF2xP9HGfZjYGuB+42t13FLqeOHVzTYHRsM6LgPnA99z9GGA3Cdod05Vwn/M5wAzgEKCSYPdEvqSt694M6P0+0gO+ATg08ngq8G6BaomdmRUThPtP3f2BcPBfcx/Zwr+NhaovBvtdU4CgR18TfoyHZK7zBqDB3Z8LH99HEPhJXtenA2+6e5O7twIPACeQ/HWd0926HVDGjfSAfwE4PPymvYTgS5mHClxTLMJ9zz8EVrn7tyKjHgIuDe9fCvxyqGuLSzfXFLgIeBI4N5wsUcsM4O4bgfVmdmQ46DRgJQle1wS7Zo43s4rwvZ5b5kSv64ju1u1DwCXh0TTHA9tzu3L6xN1H9A04G3gNeAP4X4WuJ8blPIngo9lLwPLwdjbBPukngDXh3/GFrjWm5f8Q8HB4fybwPPA68AugtND1xbC884D6cH3/JzAu6esauB54FVgB/BgoTeK6Bu4h+J6hlaCHfnl365ZgF81tYb69THCUUZ/b0qkKREQSaqTvohERkW4o4EVEEkoBLyKSUAp4EZGEUsCLiCSUAl5GFTPLmNnyyG3QfiFqZtOjZwgUKbSi3icRSZQ97j6v0EWIDAX14EUAM1tnZjea2fPh7T3h8MPM7InwXNxPmNm0cPgkM3vQzP4S3k4IZ5U2szvC85o/ZmblBVsoGfUU8DLalOftojk/Mm6Hux8H3EpwzhvC+3e7+/uBnwI3h8NvBp5y97kE54l5JRx+OHCbux8NbAM+FfPyiHRLv2SVUcXMdjOMsFAAAADmSURBVLn7mC6GrwNOdfe14UndNrp7rZltAg5299Zw+AZ3n2BmTcBUd98Xmcd04HEPLtqAmf1PoNjd/zX+JRPZn3rwIh28m/vdTdOVfZH7GfQ9lxSQAl6kw/mRv38K7z9DcCZLgIuAP4T3nwCugPZrxlYPVZEifaXehYw25Wa2PPL4UXfPHSpZambPEXR8LgyHXQksNbN/IrjK0qJw+FXAEjO7nKCnfgXBGQJFhg3tgxehfR98nbtvKnQtIoNFu2hERBJKPXgRkYRSD15EJKEU8CIiCaWAFxFJKAW8iEhCKeBFRBLq/wNXLjKu+yxXgAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "plt.plot(history.history['mse'])\n",
    "plt.plot(history.history['val_mse'])\n",
    "plt.title('Model MSE')\n",
    "plt.ylabel('MSE')\n",
    "plt.xlabel('Epoch')\n",
    "plt.legend(['Train', 'Test'], loc='upper left')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "plt.plot(history.history['mae'])\n",
    "plt.plot(history.history['val_mae'])\n",
    "plt.title('Model MAE')\n",
    "plt.ylabel('MAE')\n",
    "plt.xlabel('Epoch')\n",
    "plt.legend(['Train', 'Test'], loc='upper left')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(history.history['loss']) \n",
    "plt.plot(history.history['val_loss']) \n",
    "plt.title('Model loss') \n",
    "plt.ylabel('Loss') \n",
    "plt.xlabel('Epoch') \n",
    "plt.legend(['Train', 'Test'], loc='upper left') \n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Getting Best Metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load weights file of the best model :\n",
    "weights_file = 'neural_networks_weights.hdf5' # load the best checkpoint \n",
    "nnModel.load_weights(weights_file) # load it\n",
    "nnModel.compile(loss='mse', optimizer='adam', metrics=['mse'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Neural Networks validation MAE =  2338275.8\n"
     ]
    }
   ],
   "source": [
    "print('Neural Networks validation MAE = ', nnModel.history.history['mae'][-1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Neural Networks validation MSE =  413333730000000.0\n"
     ]
    }
   ],
   "source": [
    "print('Neural Networks validation MSE = ', nnModel.history.history['mse'][-1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Neural Networks validation RMSE =  20330610.602678906\n"
     ]
    }
   ],
   "source": [
    "print('Neural Networks validation RMSE = ', math.sqrt(nnModel.history.history['mse'][-1]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "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": "3.7.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}