dnn_tune.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import warnings\n",
    "warnings.simplefilter('ignore')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "from fast_ml.feature_selection import get_constant_features\n",
    "\n",
    "from sklearn.feature_selection import mutual_info_regression\n",
    "from sklearn.preprocessing import PowerTransformer, StandardScaler\n",
    "from statsmodels.stats.outliers_influence import variance_inflation_factor\n",
    "import scipy.stats as stats\n",
    "from sklearn.model_selection import train_test_split, cross_val_score, KFold, RandomizedSearchCV\n",
    "from sklearn.linear_model import LinearRegression, ElasticNet\n",
    "from sklearn.metrics import mean_squared_error, r2_score, mean_absolute_error\n",
    "import pickle\n",
    "import os\n",
    "from sklearn.ensemble import RandomForestRegressor\n",
    "import lightgbm\n",
    "import statsmodels.api as sm\n",
    "\n",
    "import tensorflow as tf\n",
    "from tensorflow.keras.models import Sequential\n",
    "from tensorflow.keras.layers import Dense, Dropout, BatchNormalization\n",
    "from tensorflow.keras.callbacks import EarlyStopping, ModelCheckpoint\n",
    "import keras_tuner as kt\n",
    "from keras.regularizers import l2\n",
    "\n",
    "\n",
    "\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "Xtrain = pd.read_csv('transformed_data/Xtrain_transformed.csv')\n",
    "ytrain = pd.read_csv('transformed_data/ytrain.csv')\n",
    "\n",
    "Xtest = pd.read_csv('transformed_data/Xtest_transformed.csv')\n",
    "ytest = pd.read_csv('transformed_data/ytest.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "def build_model(hyperparameter):\n",
    "    model = Sequential()\n",
    "    counter = 0\n",
    "    \n",
    "    for i in range(hyperparameter.Int('num_layers',min_value = 1, max_value = 10)):\n",
    "        if i == 0:\n",
    "            model.add(Dense(\n",
    "                hyperparameter.Int('units' + str(i), min_value = 8, max_value = 256, step = 8),\n",
    "                hyperparameter.Choice('activation' + str(i), values = ['relu', 'selu', 'tanh', 'sigmoid']),\n",
    "                kernel_initializer = 'he_uniform', kernel_regularizer = l2(0.001),\n",
    "                input_dim = Xtrain.shape[1]\n",
    "            ))\n",
    "            counter += 1\n",
    "        else:\n",
    "            model.add(Dense(\n",
    "                hyperparameter.Int('units' + str(i), min_value = 8, max_value = 256, step = 8),\n",
    "                hyperparameter.Choice('activation' + str(i), values = ['relu', 'selu', 'tanh', 'sigmoid']),\n",
    "                kernel_initializer = 'he_uniform', kernel_regularizer = l2(0.001)\n",
    "            ))\n",
    "            model.add(Dropout(hyperparameter.Choice('dropout' + str(i), values=[0.1,0.2,0.3,0.4,0.5,0.6])))\n",
    "            model.add(BatchNormalization())\n",
    "            \n",
    "    model.add(Dense(1, activation='linear'))\n",
    "    model.compile(hyperparameter.Choice(\n",
    "        'optimizer', values = ['adam', 'rmsprop', 'sgd', 'adadelta', 'adagrad']), \n",
    "                  loss='mean_squared_error', metrics=['mse']\n",
    "                  )\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "tuner = kt.RandomSearch(hypermodel = build_model, max_trials = 5, objective = ['val_mse'], \n",
    "                     directory = 'DNN_RandomSearch_Tuner', project_name = 'trial_03')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Trial 5 Complete [00h 01m 33s]\n",
      "multi_objective: nan\n",
      "\n",
      "Best multi_objective So Far: 17405896704.0\n",
      "Total elapsed time: 00h 11m 08s\n"
     ]
    }
   ],
   "source": [
    "tuner.search(Xtrain, ytrain, epochs = 5, validation_data = (Xtest, ytest), validation_split = 0.3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " dense (Dense)               (None, 224)               2912      \n",
      "                                                                 \n",
      " dense_1 (Dense)             (None, 208)               46800     \n",
      "                                                                 \n",
      " dropout (Dropout)           (None, 208)               0         \n",
      "                                                                 \n",
      " batch_normalization (BatchN  (None, 208)              832       \n",
      " ormalization)                                                   \n",
      "                                                                 \n",
      " dense_2 (Dense)             (None, 224)               46816     \n",
      "                                                                 \n",
      " dropout_1 (Dropout)         (None, 224)               0         \n",
      "                                                                 \n",
      " batch_normalization_1 (Batc  (None, 224)              896       \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " dense_3 (Dense)             (None, 120)               27000     \n",
      "                                                                 \n",
      " dropout_2 (Dropout)         (None, 120)               0         \n",
      "                                                                 \n",
      " batch_normalization_2 (Batc  (None, 120)              480       \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " dense_4 (Dense)             (None, 104)               12584     \n",
      "                                                                 \n",
      " dropout_3 (Dropout)         (None, 104)               0         \n",
      "                                                                 \n",
      " batch_normalization_3 (Batc  (None, 104)              416       \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " dense_5 (Dense)             (None, 224)               23520     \n",
      "                                                                 \n",
      " dropout_4 (Dropout)         (None, 224)               0         \n",
      "                                                                 \n",
      " batch_normalization_4 (Batc  (None, 224)              896       \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " dense_6 (Dense)             (None, 232)               52200     \n",
      "                                                                 \n",
      " dropout_5 (Dropout)         (None, 232)               0         \n",
      "                                                                 \n",
      " batch_normalization_5 (Batc  (None, 232)              928       \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " dense_7 (Dense)             (None, 136)               31688     \n",
      "                                                                 \n",
      " dropout_6 (Dropout)         (None, 136)               0         \n",
      "                                                                 \n",
      " batch_normalization_6 (Batc  (None, 136)              544       \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " dense_8 (Dense)             (None, 1)                 137       \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 248,649\n",
      "Trainable params: 246,153\n",
      "Non-trainable params: 2,496\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "tuned_model = tuner.get_best_models()[0]\n",
    "tuned_model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "early_stopping = EarlyStopping(\n",
    "    monitor='val_mse',\n",
    "    min_delta=0.003,\n",
    "    patience=5,\n",
    "    verbose=1,\n",
    "    mode='auto',\n",
    "    baseline=None,\n",
    "    restore_best_weights=True\n",
    ")\n",
    "\n",
    "model_checkpoint = ModelCheckpoint(\n",
    "    'dnn_tuned.h5',\n",
    "    monitor='val_mse',\n",
    "    mode='auto',\n",
    "    verbose=1,\n",
    "    save_best_only=True\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/50\n",
      "2233/2235 [============================>.] - ETA: 0s - loss: 17905082368.0000 - mse: 17905082368.0000\n",
      "Epoch 1: val_mse improved from inf to 17405310976.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 31s 14ms/step - loss: 17903097856.0000 - mse: 17903097856.0000 - val_loss: 17405310976.0000 - val_mse: 17405310976.0000\n",
      "Epoch 2/50\n",
      "2233/2235 [============================>.] - ETA: 0s - loss: 17904781312.0000 - mse: 17904781312.0000\n",
      "Epoch 2: val_mse improved from 17405310976.00000 to 17404547072.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 30s 13ms/step - loss: 17902661632.0000 - mse: 17902661632.0000 - val_loss: 17404547072.0000 - val_mse: 17404547072.0000\n",
      "Epoch 3/50\n",
      "2233/2235 [============================>.] - ETA: 0s - loss: 17897119744.0000 - mse: 17897119744.0000\n",
      "Epoch 3: val_mse improved from 17404547072.00000 to 17403719680.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 30s 13ms/step - loss: 17902084096.0000 - mse: 17902084096.0000 - val_loss: 17403719680.0000 - val_mse: 17403719680.0000\n",
      "Epoch 4/50\n",
      "2234/2235 [============================>.] - ETA: 0s - loss: 17904142336.0000 - mse: 17904142336.0000\n",
      "Epoch 4: val_mse improved from 17403719680.00000 to 17402867712.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 30s 13ms/step - loss: 17901402112.0000 - mse: 17901402112.0000 - val_loss: 17402867712.0000 - val_mse: 17402867712.0000\n",
      "Epoch 5/50\n",
      "2233/2235 [============================>.] - ETA: 0s - loss: 17900859392.0000 - mse: 17900859392.0000\n",
      "Epoch 5: val_mse improved from 17402867712.00000 to 17402234880.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 30s 13ms/step - loss: 17900705792.0000 - mse: 17900705792.0000 - val_loss: 17402234880.0000 - val_mse: 17402234880.0000\n",
      "Epoch 6/50\n",
      "2231/2235 [============================>.] - ETA: 0s - loss: 17900826624.0000 - mse: 17900826624.0000\n",
      "Epoch 6: val_mse improved from 17402234880.00000 to 17401538560.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 30s 13ms/step - loss: 17900046336.0000 - mse: 17900046336.0000 - val_loss: 17401538560.0000 - val_mse: 17401538560.0000\n",
      "Epoch 7/50\n",
      "2231/2235 [============================>.] - ETA: 0s - loss: 17898487808.0000 - mse: 17898487808.0000\n",
      "Epoch 7: val_mse improved from 17401538560.00000 to 17400928256.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 30s 13ms/step - loss: 17899440128.0000 - mse: 17899440128.0000 - val_loss: 17400928256.0000 - val_mse: 17400928256.0000\n",
      "Epoch 8/50\n",
      "2235/2235 [==============================] - ETA: 0s - loss: 17898872832.0000 - mse: 17898872832.0000\n",
      "Epoch 8: val_mse improved from 17400928256.00000 to 17400360960.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 30s 13ms/step - loss: 17898872832.0000 - mse: 17898872832.0000 - val_loss: 17400360960.0000 - val_mse: 17400360960.0000\n",
      "Epoch 9/50\n",
      "2233/2235 [============================>.] - ETA: 0s - loss: 17905203200.0000 - mse: 17905203200.0000\n",
      "Epoch 9: val_mse improved from 17400360960.00000 to 17399840768.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 30s 13ms/step - loss: 17898295296.0000 - mse: 17898295296.0000 - val_loss: 17399840768.0000 - val_mse: 17399840768.0000\n",
      "Epoch 10/50\n",
      "2233/2235 [============================>.] - ETA: 0s - loss: 17896896512.0000 - mse: 17896896512.0000\n",
      "Epoch 10: val_mse improved from 17399840768.00000 to 17399166976.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 30s 14ms/step - loss: 17897697280.0000 - mse: 17897697280.0000 - val_loss: 17399166976.0000 - val_mse: 17399166976.0000\n",
      "Epoch 11/50\n",
      "2234/2235 [============================>.] - ETA: 0s - loss: 17898520576.0000 - mse: 17898520576.0000\n",
      "Epoch 11: val_mse improved from 17399166976.00000 to 17398657024.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 30s 13ms/step - loss: 17897191424.0000 - mse: 17897191424.0000 - val_loss: 17398657024.0000 - val_mse: 17398657024.0000\n",
      "Epoch 12/50\n",
      "2232/2235 [============================>.] - ETA: 0s - loss: 17907562496.0000 - mse: 17907562496.0000\n",
      "Epoch 12: val_mse improved from 17398657024.00000 to 17398083584.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 30s 13ms/step - loss: 17896615936.0000 - mse: 17896615936.0000 - val_loss: 17398083584.0000 - val_mse: 17398083584.0000\n",
      "Epoch 13/50\n",
      "2234/2235 [============================>.] - ETA: 0s - loss: 17899063296.0000 - mse: 17899063296.0000\n",
      "Epoch 13: val_mse improved from 17398083584.00000 to 17397458944.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 30s 13ms/step - loss: 17896046592.0000 - mse: 17896046592.0000 - val_loss: 17397458944.0000 - val_mse: 17397458944.0000\n",
      "Epoch 14/50\n",
      "2234/2235 [============================>.] - ETA: 0s - loss: 17897723904.0000 - mse: 17897723904.0000\n",
      "Epoch 14: val_mse improved from 17397458944.00000 to 17396744192.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 30s 13ms/step - loss: 17895501824.0000 - mse: 17895501824.0000 - val_loss: 17396744192.0000 - val_mse: 17396744192.0000\n",
      "Epoch 15/50\n",
      "2233/2235 [============================>.] - ETA: 0s - loss: 17898297344.0000 - mse: 17898297344.0000\n",
      "Epoch 15: val_mse improved from 17396744192.00000 to 17396221952.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 30s 13ms/step - loss: 17894950912.0000 - mse: 17894950912.0000 - val_loss: 17396221952.0000 - val_mse: 17396221952.0000\n",
      "Epoch 16/50\n",
      "2232/2235 [============================>.] - ETA: 0s - loss: 17893545984.0000 - mse: 17893545984.0000\n",
      "Epoch 16: val_mse improved from 17396221952.00000 to 17395687424.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 30s 13ms/step - loss: 17894436864.0000 - mse: 17894436864.0000 - val_loss: 17395687424.0000 - val_mse: 17395687424.0000\n",
      "Epoch 17/50\n",
      "2235/2235 [==============================] - ETA: 0s - loss: 17893865472.0000 - mse: 17893865472.0000\n",
      "Epoch 17: val_mse improved from 17395687424.00000 to 17395189760.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 30s 13ms/step - loss: 17893865472.0000 - mse: 17893865472.0000 - val_loss: 17395189760.0000 - val_mse: 17395189760.0000\n",
      "Epoch 18/50\n",
      "2232/2235 [============================>.] - ETA: 0s - loss: 17898393600.0000 - mse: 17898393600.0000\n",
      "Epoch 18: val_mse improved from 17395189760.00000 to 17394804736.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 30s 13ms/step - loss: 17893326848.0000 - mse: 17893326848.0000 - val_loss: 17394804736.0000 - val_mse: 17394804736.0000\n",
      "Epoch 19/50\n",
      "2233/2235 [============================>.] - ETA: 0s - loss: 17885542400.0000 - mse: 17885542400.0000\n",
      "Epoch 19: val_mse improved from 17394804736.00000 to 17393920000.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 30s 13ms/step - loss: 17892773888.0000 - mse: 17892773888.0000 - val_loss: 17393920000.0000 - val_mse: 17393920000.0000\n",
      "Epoch 20/50\n",
      "2235/2235 [==============================] - ETA: 0s - loss: 17892229120.0000 - mse: 17892229120.0000\n",
      "Epoch 20: val_mse improved from 17393920000.00000 to 17393453056.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 30s 13ms/step - loss: 17892229120.0000 - mse: 17892229120.0000 - val_loss: 17393453056.0000 - val_mse: 17393453056.0000\n",
      "Epoch 21/50\n",
      "2232/2235 [============================>.] - ETA: 0s - loss: 17890222080.0000 - mse: 17890222080.0000\n",
      "Epoch 21: val_mse improved from 17393453056.00000 to 17393197056.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 37s 17ms/step - loss: 17891670016.0000 - mse: 17891670016.0000 - val_loss: 17393197056.0000 - val_mse: 17393197056.0000\n",
      "Epoch 22/50\n",
      "2232/2235 [============================>.] - ETA: 0s - loss: 17899558912.0000 - mse: 17899558912.0000\n",
      "Epoch 22: val_mse improved from 17393197056.00000 to 17392625664.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 31s 14ms/step - loss: 17891151872.0000 - mse: 17891151872.0000 - val_loss: 17392625664.0000 - val_mse: 17392625664.0000\n",
      "Epoch 23/50\n",
      "2233/2235 [============================>.] - ETA: 0s - loss: 17885038592.0000 - mse: 17885038592.0000\n",
      "Epoch 23: val_mse improved from 17392625664.00000 to 17392119808.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 31s 14ms/step - loss: 17890625536.0000 - mse: 17890625536.0000 - val_loss: 17392119808.0000 - val_mse: 17392119808.0000\n",
      "Epoch 24/50\n",
      "2234/2235 [============================>.] - ETA: 0s - loss: 17890560000.0000 - mse: 17890560000.0000\n",
      "Epoch 24: val_mse improved from 17392119808.00000 to 17391355904.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 31s 14ms/step - loss: 17890050048.0000 - mse: 17890050048.0000 - val_loss: 17391355904.0000 - val_mse: 17391355904.0000\n",
      "Epoch 25/50\n",
      "2234/2235 [============================>.] - ETA: 0s - loss: 17889347584.0000 - mse: 17889347584.0000\n",
      "Epoch 25: val_mse improved from 17391355904.00000 to 17390868480.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 34s 15ms/step - loss: 17889546240.0000 - mse: 17889546240.0000 - val_loss: 17390868480.0000 - val_mse: 17390868480.0000\n",
      "Epoch 26/50\n",
      "2232/2235 [============================>.] - ETA: 0s - loss: 17885081600.0000 - mse: 17885081600.0000\n",
      "Epoch 26: val_mse improved from 17390868480.00000 to 17390348288.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 30s 14ms/step - loss: 17889036288.0000 - mse: 17889036288.0000 - val_loss: 17390348288.0000 - val_mse: 17390348288.0000\n",
      "Epoch 27/50\n",
      "2233/2235 [============================>.] - ETA: 0s - loss: 17883910144.0000 - mse: 17883910144.0000\n",
      "Epoch 27: val_mse improved from 17390348288.00000 to 17389592576.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 31s 14ms/step - loss: 17888458752.0000 - mse: 17888458752.0000 - val_loss: 17389592576.0000 - val_mse: 17389592576.0000\n",
      "Epoch 28/50\n",
      "2234/2235 [============================>.] - ETA: 0s - loss: 17881833472.0000 - mse: 17881833472.0000\n",
      "Epoch 28: val_mse improved from 17389592576.00000 to 17389205504.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 31s 14ms/step - loss: 17887967232.0000 - mse: 17887967232.0000 - val_loss: 17389205504.0000 - val_mse: 17389205504.0000\n",
      "Epoch 29/50\n",
      "2235/2235 [==============================] - ETA: 0s - loss: 17887397888.0000 - mse: 17887397888.0000\n",
      "Epoch 29: val_mse improved from 17389205504.00000 to 17388617728.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 30s 14ms/step - loss: 17887397888.0000 - mse: 17887397888.0000 - val_loss: 17388617728.0000 - val_mse: 17388617728.0000\n",
      "Epoch 30/50\n",
      "2233/2235 [============================>.] - ETA: 0s - loss: 17889882112.0000 - mse: 17889882112.0000\n",
      "Epoch 30: val_mse improved from 17388617728.00000 to 17388103680.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 30s 14ms/step - loss: 17886877696.0000 - mse: 17886877696.0000 - val_loss: 17388103680.0000 - val_mse: 17388103680.0000\n",
      "Epoch 31/50\n",
      "2235/2235 [==============================] - ETA: 0s - loss: 17886375936.0000 - mse: 17886375936.0000\n",
      "Epoch 31: val_mse improved from 17388103680.00000 to 17387614208.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 31s 14ms/step - loss: 17886375936.0000 - mse: 17886375936.0000 - val_loss: 17387614208.0000 - val_mse: 17387614208.0000\n",
      "Epoch 32/50\n",
      "2235/2235 [==============================] - ETA: 0s - loss: 17885802496.0000 - mse: 17885802496.0000\n",
      "Epoch 32: val_mse improved from 17387614208.00000 to 17387032576.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 31s 14ms/step - loss: 17885802496.0000 - mse: 17885802496.0000 - val_loss: 17387032576.0000 - val_mse: 17387032576.0000\n",
      "Epoch 33/50\n",
      "2235/2235 [==============================] - ETA: 0s - loss: 17885255680.0000 - mse: 17885255680.0000\n",
      "Epoch 33: val_mse improved from 17387032576.00000 to 17386434560.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 31s 14ms/step - loss: 17885255680.0000 - mse: 17885255680.0000 - val_loss: 17386434560.0000 - val_mse: 17386434560.0000\n",
      "Epoch 34/50\n",
      "2231/2235 [============================>.] - ETA: 0s - loss: 17890379776.0000 - mse: 17890379776.0000\n",
      "Epoch 34: val_mse improved from 17386434560.00000 to 17385885696.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 31s 14ms/step - loss: 17884753920.0000 - mse: 17884753920.0000 - val_loss: 17385885696.0000 - val_mse: 17385885696.0000\n",
      "Epoch 35/50\n",
      "2233/2235 [============================>.] - ETA: 0s - loss: 17888782336.0000 - mse: 17888782336.0000\n",
      "Epoch 35: val_mse improved from 17385885696.00000 to 17385623552.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 31s 14ms/step - loss: 17884211200.0000 - mse: 17884211200.0000 - val_loss: 17385623552.0000 - val_mse: 17385623552.0000\n",
      "Epoch 36/50\n",
      "2233/2235 [============================>.] - ETA: 0s - loss: 17889495040.0000 - mse: 17889495040.0000\n",
      "Epoch 36: val_mse improved from 17385623552.00000 to 17385109504.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 31s 14ms/step - loss: 17883693056.0000 - mse: 17883693056.0000 - val_loss: 17385109504.0000 - val_mse: 17385109504.0000\n",
      "Epoch 37/50\n",
      "2235/2235 [==============================] - ETA: 0s - loss: 17883109376.0000 - mse: 17883109376.0000\n",
      "Epoch 37: val_mse improved from 17385109504.00000 to 17384267776.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 31s 14ms/step - loss: 17883109376.0000 - mse: 17883109376.0000 - val_loss: 17384267776.0000 - val_mse: 17384267776.0000\n",
      "Epoch 38/50\n",
      "2233/2235 [============================>.] - ETA: 0s - loss: 17873590272.0000 - mse: 17873590272.0000\n",
      "Epoch 38: val_mse improved from 17384267776.00000 to 17384136704.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 31s 14ms/step - loss: 17882621952.0000 - mse: 17882621952.0000 - val_loss: 17384136704.0000 - val_mse: 17384136704.0000\n",
      "Epoch 39/50\n",
      "2234/2235 [============================>.] - ETA: 0s - loss: 17881790464.0000 - mse: 17881790464.0000\n",
      "Epoch 39: val_mse improved from 17384136704.00000 to 17383614464.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 31s 14ms/step - loss: 17882107904.0000 - mse: 17882107904.0000 - val_loss: 17383614464.0000 - val_mse: 17383614464.0000\n",
      "Epoch 40/50\n",
      "2234/2235 [============================>.] - ETA: 0s - loss: 17885763584.0000 - mse: 17885763584.0000\n",
      "Epoch 40: val_mse improved from 17383614464.00000 to 17382823936.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 31s 14ms/step - loss: 17881556992.0000 - mse: 17881556992.0000 - val_loss: 17382823936.0000 - val_mse: 17382823936.0000\n",
      "Epoch 41/50\n",
      "2235/2235 [==============================] - ETA: 0s - loss: 17881014272.0000 - mse: 17881014272.0000\n",
      "Epoch 41: val_mse improved from 17382823936.00000 to 17382072320.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 31s 14ms/step - loss: 17881014272.0000 - mse: 17881014272.0000 - val_loss: 17382072320.0000 - val_mse: 17382072320.0000\n",
      "Epoch 42/50\n",
      "2235/2235 [==============================] - ETA: 0s - loss: 17880492032.0000 - mse: 17880492032.0000\n",
      "Epoch 42: val_mse improved from 17382072320.00000 to 17381574656.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 31s 14ms/step - loss: 17880492032.0000 - mse: 17880492032.0000 - val_loss: 17381574656.0000 - val_mse: 17381574656.0000\n",
      "Epoch 43/50\n",
      "2235/2235 [==============================] - ETA: 0s - loss: 17879926784.0000 - mse: 17879926784.0000\n",
      "Epoch 43: val_mse improved from 17381574656.00000 to 17381183488.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 31s 14ms/step - loss: 17879926784.0000 - mse: 17879926784.0000 - val_loss: 17381183488.0000 - val_mse: 17381183488.0000\n",
      "Epoch 44/50\n",
      "2235/2235 [==============================] - ETA: 0s - loss: 17879402496.0000 - mse: 17879402496.0000\n",
      "Epoch 44: val_mse improved from 17381183488.00000 to 17380560896.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 31s 14ms/step - loss: 17879402496.0000 - mse: 17879402496.0000 - val_loss: 17380560896.0000 - val_mse: 17380560896.0000\n",
      "Epoch 45/50\n",
      "2234/2235 [============================>.] - ETA: 0s - loss: 17877950464.0000 - mse: 17877950464.0000\n",
      "Epoch 45: val_mse improved from 17380560896.00000 to 17380184064.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 31s 14ms/step - loss: 17878894592.0000 - mse: 17878894592.0000 - val_loss: 17380184064.0000 - val_mse: 17380184064.0000\n",
      "Epoch 46/50\n",
      "2233/2235 [============================>.] - ETA: 0s - loss: 17876385792.0000 - mse: 17876385792.0000\n",
      "Epoch 46: val_mse improved from 17380184064.00000 to 17379645440.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 31s 14ms/step - loss: 17878382592.0000 - mse: 17878382592.0000 - val_loss: 17379645440.0000 - val_mse: 17379645440.0000\n",
      "Epoch 47/50\n",
      "2234/2235 [============================>.] - ETA: 0s - loss: 17881321472.0000 - mse: 17881321472.0000\n",
      "Epoch 47: val_mse improved from 17379645440.00000 to 17378992128.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 31s 14ms/step - loss: 17877854208.0000 - mse: 17877854208.0000 - val_loss: 17378992128.0000 - val_mse: 17378992128.0000\n",
      "Epoch 48/50\n",
      "2233/2235 [============================>.] - ETA: 0s - loss: 17874145280.0000 - mse: 17874145280.0000\n",
      "Epoch 48: val_mse improved from 17378992128.00000 to 17378658304.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 31s 14ms/step - loss: 17877309440.0000 - mse: 17877309440.0000 - val_loss: 17378658304.0000 - val_mse: 17378658304.0000\n",
      "Epoch 49/50\n",
      "2233/2235 [============================>.] - ETA: 0s - loss: 17878401024.0000 - mse: 17878401024.0000\n",
      "Epoch 49: val_mse improved from 17378658304.00000 to 17378357248.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 31s 14ms/step - loss: 17876776960.0000 - mse: 17876776960.0000 - val_loss: 17378357248.0000 - val_mse: 17378357248.0000\n",
      "Epoch 50/50\n",
      "2233/2235 [============================>.] - ETA: 0s - loss: 17880418304.0000 - mse: 17880418304.0000\n",
      "Epoch 50: val_mse improved from 17378357248.00000 to 17377642496.00000, saving model to dnn_tuned.h5\n",
      "2235/2235 [==============================] - 31s 14ms/step - loss: 17876246528.0000 - mse: 17876246528.0000 - val_loss: 17377642496.0000 - val_mse: 17377642496.0000\n"
     ]
    }
   ],
   "source": [
    "history = tuned_model.fit(Xtrain, ytrain, batch_size=32, epochs=50, \n",
    "                    validation_data=(Xtest, ytest),\n",
    "                    validation_split=0.2,\n",
    "                    callbacks=[early_stopping, model_checkpoint])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 300x300 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(3,3))\n",
    "\n",
    "plt.plot(history.history['val_mse'], color='blue', label='Validation MSE')\n",
    "plt.plot(history.history['mse'], color='orange', label='Training MSE')\n",
    "\n",
    "plt.xlabel('Epoch')\n",
    "plt.ylabel('Mean Squared Error')\n",
    "plt.title('Training and Validation Mean Squared Error')\n",
    "\n",
    "plt.legend()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 300x300 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(3,3))\n",
    "\n",
    "plt.plot(history.history['val_loss'], color='blue', label='Validation Loss')\n",
    "plt.plot(history.history['loss'], color='orange', label='Training Loss')\n",
    "\n",
    "plt.xlabel('Epoch')\n",
    "plt.ylabel('Loss')\n",
    "plt.title('Training and Validation Loss')\n",
    "\n",
    "plt.legend()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "dict_keys(['loss', 'mse', 'val_loss', 'val_mse'])"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "history.history.keys()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "559/559 [==============================] - 2s 3ms/step\n",
      "R2 Score: -1.1716981337871362\n"
     ]
    }
   ],
   "source": [
    "y_pred = tuned_model.predict(Xtest)\n",
    "\n",
    "r2_score_dnn_tuned = r2_score(ytest, y_pred)\n",
    "print('R2 Score:', r2_score_dnn_tuned)"
   ]
  },
  {
   "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": "3.9.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}