temp-predictions(time-series-LSTM).ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "from sklearn.preprocessing import StandardScaler"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.read_csv('temp.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>No</th>\n",
       "      <th>year</th>\n",
       "      <th>month</th>\n",
       "      <th>day</th>\n",
       "      <th>hour</th>\n",
       "      <th>pm2.5</th>\n",
       "      <th>DEWP</th>\n",
       "      <th>TEMP</th>\n",
       "      <th>PRES</th>\n",
       "      <th>cbwd</th>\n",
       "      <th>Iws</th>\n",
       "      <th>Is</th>\n",
       "      <th>Ir</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>2010</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-21</td>\n",
       "      <td>-11.0</td>\n",
       "      <td>1021.0</td>\n",
       "      <td>NW</td>\n",
       "      <td>1.79</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2</td>\n",
       "      <td>2010</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-21</td>\n",
       "      <td>-12.0</td>\n",
       "      <td>1020.0</td>\n",
       "      <td>NW</td>\n",
       "      <td>4.92</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>2010</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-21</td>\n",
       "      <td>-11.0</td>\n",
       "      <td>1019.0</td>\n",
       "      <td>NW</td>\n",
       "      <td>6.71</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4</td>\n",
       "      <td>2010</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-21</td>\n",
       "      <td>-14.0</td>\n",
       "      <td>1019.0</td>\n",
       "      <td>NW</td>\n",
       "      <td>9.84</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5</td>\n",
       "      <td>2010</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>4</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-20</td>\n",
       "      <td>-12.0</td>\n",
       "      <td>1018.0</td>\n",
       "      <td>NW</td>\n",
       "      <td>12.97</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>6</td>\n",
       "      <td>2010</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-19</td>\n",
       "      <td>-10.0</td>\n",
       "      <td>1017.0</td>\n",
       "      <td>NW</td>\n",
       "      <td>16.10</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>7</td>\n",
       "      <td>2010</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>6</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-19</td>\n",
       "      <td>-9.0</td>\n",
       "      <td>1017.0</td>\n",
       "      <td>NW</td>\n",
       "      <td>19.23</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>8</td>\n",
       "      <td>2010</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>7</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-19</td>\n",
       "      <td>-9.0</td>\n",
       "      <td>1017.0</td>\n",
       "      <td>NW</td>\n",
       "      <td>21.02</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>9</td>\n",
       "      <td>2010</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>8</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-19</td>\n",
       "      <td>-9.0</td>\n",
       "      <td>1017.0</td>\n",
       "      <td>NW</td>\n",
       "      <td>24.15</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>10</td>\n",
       "      <td>2010</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>9</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-20</td>\n",
       "      <td>-8.0</td>\n",
       "      <td>1017.0</td>\n",
       "      <td>NW</td>\n",
       "      <td>27.28</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   No  year  month  day  hour  pm2.5  DEWP  TEMP    PRES cbwd    Iws  Is  Ir\n",
       "0   1  2010      1    1     0    NaN   -21 -11.0  1021.0   NW   1.79   0   0\n",
       "1   2  2010      1    1     1    NaN   -21 -12.0  1020.0   NW   4.92   0   0\n",
       "2   3  2010      1    1     2    NaN   -21 -11.0  1019.0   NW   6.71   0   0\n",
       "3   4  2010      1    1     3    NaN   -21 -14.0  1019.0   NW   9.84   0   0\n",
       "4   5  2010      1    1     4    NaN   -20 -12.0  1018.0   NW  12.97   0   0\n",
       "5   6  2010      1    1     5    NaN   -19 -10.0  1017.0   NW  16.10   0   0\n",
       "6   7  2010      1    1     6    NaN   -19  -9.0  1017.0   NW  19.23   0   0\n",
       "7   8  2010      1    1     7    NaN   -19  -9.0  1017.0   NW  21.02   0   0\n",
       "8   9  2010      1    1     8    NaN   -19  -9.0  1017.0   NW  24.15   0   0\n",
       "9  10  2010      1    1     9    NaN   -20  -8.0  1017.0   NW  27.28   0   0"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['No', 'year', 'month', 'day', 'hour', 'pm2.5', 'DEWP', 'TEMP', 'PRES',\n",
       "       'cbwd', 'Iws', 'Is', 'Ir'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset = df[['year','month','day','hour','DEWP','TEMP','PRES','Iws','Is','Ir']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "from collections import Counter\n",
    "cut = int(dataset.shape[0]*0.8)\n",
    "train = dataset.loc[:cut]\n",
    "test = dataset.loc[cut:]\n",
    "\n",
    "train = train.drop(['year'], axis=1)\n",
    "test = test.drop(['year'], axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "for s in range(1,25):\n",
    "    train['shift_{}'.format(s)] = train['TEMP'].shift(s)\n",
    "    test['shift_{}'.format(s)] = test['TEMP'].shift(s)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train = train.dropna().drop('TEMP', axis=1)\n",
    "y_train = train.dropna()[['TEMP']]\n",
    "\n",
    "X_test = test.dropna().drop('TEMP', axis=1)\n",
    "y_test = test.dropna()[['TEMP']]\n",
    "\n",
    "X_train_sc = StandardScaler().fit_transform(X_train)\n",
    "y_train_sc = StandardScaler().fit_transform(y_train)\n",
    "\n",
    "X_test_sc = StandardScaler().fit_transform(X_test)\n",
    "y_test_sc = StandardScaler().fit_transform(y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "train.to_csv('train.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/fatmakursun/anaconda3/lib/python3.6/site-packages/h5py/__init__.py:36: FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. In future, it will be treated as `np.float64 == np.dtype(float).type`.\n",
      "  from ._conv import register_converters as _register_converters\n",
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "from keras.models import Sequential\n",
    "from keras.layers import Dense\n",
    "from keras.layers import Dropout\n",
    "from keras.layers import LSTM\n",
    "from keras.optimizers import Adam\n",
    "from keras import regularizers\n",
    "import keras.backend as K"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Hyperparameter for Deep Learning\n",
    "learning_rate = 0.01\n",
    "training_epochs = 200\n",
    "display_step = 200\n",
    "factor = 0.10\n",
    "batch = 128\n",
    "seed = 7\n",
    "beta_1 = 0.9\n",
    "beta_2 = 0.999\n",
    "epsilon = None\n",
    "decay = 0.001\n",
    "np.random.seed(seed)\n",
    "dims = X_train_sc.shape[1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense_1 (Dense)              (None, 128)               4224      \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 128)               0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 64)                8256      \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 64)                0         \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 32)                2080      \n",
      "_________________________________________________________________\n",
      "dropout_3 (Dropout)          (None, 32)                0         \n",
      "_________________________________________________________________\n",
      "dense_4 (Dense)              (None, 1)                 33        \n",
      "=================================================================\n",
      "Total params: 14,593\n",
      "Trainable params: 14,593\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model = Sequential()\n",
    "model.add(Dense(128,\n",
    "                input_shape=(dims,),\n",
    "                activation='relu',\n",
    "                kernel_regularizer=regularizers.l2(0.01)))\n",
    "model.add(Dropout(0.3))\n",
    "model.add(Dense(64, activation='relu'))\n",
    "model.add(Dropout(0.3))\n",
    "model.add(Dense(32, activation='relu'))\n",
    "model.add(Dropout(0.3))\n",
    "model.add(Dense(1))\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 35036 samples, validate on 8741 samples\n",
      "Epoch 1/200\n",
      "35036/35036 [==============================] - 2s 48us/step - loss: 0.1928 - val_loss: 0.0557\n",
      "Epoch 2/200\n",
      "35036/35036 [==============================] - 1s 30us/step - loss: 0.0948 - val_loss: 0.0573\n",
      "Epoch 3/200\n",
      "35036/35036 [==============================] - 1s 30us/step - loss: 0.0933 - val_loss: 0.0506\n",
      "Epoch 4/200\n",
      "35036/35036 [==============================] - 1s 30us/step - loss: 0.0936 - val_loss: 0.0620\n",
      "Epoch 5/200\n",
      "35036/35036 [==============================] - 1s 30us/step - loss: 0.0936 - val_loss: 0.0795\n",
      "Epoch 6/200\n",
      "35036/35036 [==============================] - 1s 30us/step - loss: 0.0972 - val_loss: 0.0627\n",
      "Epoch 7/200\n",
      "35036/35036 [==============================] - 1s 31us/step - loss: 0.0931 - val_loss: 0.0509\n",
      "Epoch 8/200\n",
      "35036/35036 [==============================] - 1s 30us/step - loss: 0.0949 - val_loss: 0.0555\n",
      "Epoch 9/200\n",
      "35036/35036 [==============================] - 1s 29us/step - loss: 0.0980 - val_loss: 0.0860\n",
      "Epoch 10/200\n",
      "35036/35036 [==============================] - 1s 29us/step - loss: 0.0977 - val_loss: 0.0608\n",
      "Epoch 11/200\n",
      "35036/35036 [==============================] - 1s 29us/step - loss: 0.0970 - val_loss: 0.0736\n",
      "Epoch 12/200\n",
      "35036/35036 [==============================] - 1s 29us/step - loss: 0.0987 - val_loss: 0.0687\n",
      "Epoch 13/200\n",
      "35036/35036 [==============================] - 1s 30us/step - loss: 0.0941 - val_loss: 0.0962\n",
      "Epoch 14/200\n",
      "35036/35036 [==============================] - 1s 29us/step - loss: 0.0963 - val_loss: 0.0782\n",
      "Epoch 15/200\n",
      "35036/35036 [==============================] - 1s 31us/step - loss: 0.0970 - val_loss: 0.1031\n",
      "Epoch 16/200\n",
      "35036/35036 [==============================] - 1s 30us/step - loss: 0.0963 - val_loss: 0.0974\n",
      "Epoch 17/200\n",
      "35036/35036 [==============================] - 1s 30us/step - loss: 0.0937 - val_loss: 0.0770\n",
      "Epoch 18/200\n",
      "35036/35036 [==============================] - 1s 30us/step - loss: 0.0941 - val_loss: 0.0723\n",
      "Epoch 19/200\n",
      "35036/35036 [==============================] - 1s 30us/step - loss: 0.0941 - val_loss: 0.1043\n",
      "Epoch 20/200\n",
      "35036/35036 [==============================] - 1s 30us/step - loss: 0.0922 - val_loss: 0.0706\n",
      "Epoch 21/200\n",
      "35036/35036 [==============================] - 1s 30us/step - loss: 0.0908 - val_loss: 0.0922\n",
      "Epoch 22/200\n",
      "35036/35036 [==============================] - 1s 30us/step - loss: 0.0923 - val_loss: 0.0491\n",
      "Epoch 23/200\n",
      "35036/35036 [==============================] - 1s 31us/step - loss: 0.0906 - val_loss: 0.0719\n",
      "Epoch 24/200\n",
      "35036/35036 [==============================] - 1s 32us/step - loss: 0.0919 - val_loss: 0.1199\n",
      "Epoch 25/200\n",
      "35036/35036 [==============================] - 1s 31us/step - loss: 0.0926 - val_loss: 0.1213\n",
      "Epoch 26/200\n",
      "35036/35036 [==============================] - 1s 32us/step - loss: 0.0890 - val_loss: 0.1014\n",
      "Epoch 27/200\n",
      "35036/35036 [==============================] - 1s 31us/step - loss: 0.0936 - val_loss: 0.0887\n",
      "Epoch 28/200\n",
      "35036/35036 [==============================] - 1s 32us/step - loss: 0.0879 - val_loss: 0.0989\n",
      "Epoch 29/200\n",
      "35036/35036 [==============================] - 1s 31us/step - loss: 0.0893 - val_loss: 0.1031\n",
      "Epoch 30/200\n",
      "35036/35036 [==============================] - 1s 30us/step - loss: 0.0888 - val_loss: 0.0831\n",
      "Epoch 31/200\n",
      "35036/35036 [==============================] - 1s 30us/step - loss: 0.0889 - val_loss: 0.1426\n",
      "Epoch 32/200\n",
      "35036/35036 [==============================] - 1s 30us/step - loss: 0.0910 - val_loss: 0.1328\n",
      "Epoch 33/200\n",
      "35036/35036 [==============================] - 1s 30us/step - loss: 0.0934 - val_loss: 0.1295\n",
      "Epoch 34/200\n",
      "35036/35036 [==============================] - 1s 30us/step - loss: 0.0917 - val_loss: 0.0714\n",
      "Epoch 35/200\n",
      "35036/35036 [==============================] - 1s 32us/step - loss: 0.0860 - val_loss: 0.1016\n",
      "Epoch 36/200\n",
      "35036/35036 [==============================] - 1s 30us/step - loss: 0.0878 - val_loss: 0.1173\n",
      "Epoch 37/200\n",
      "35036/35036 [==============================] - 1s 31us/step - loss: 0.0906 - val_loss: 0.1259\n",
      "Epoch 38/200\n",
      "35036/35036 [==============================] - 1s 32us/step - loss: 0.0908 - val_loss: 0.1046\n",
      "Epoch 39/200\n",
      "35036/35036 [==============================] - 1s 32us/step - loss: 0.0888 - val_loss: 0.1011\n",
      "Epoch 40/200\n",
      "35036/35036 [==============================] - 1s 32us/step - loss: 0.0872 - val_loss: 0.1221\n",
      "Epoch 41/200\n",
      "35036/35036 [==============================] - 1s 33us/step - loss: 0.0896 - val_loss: 0.1116\n",
      "Epoch 42/200\n",
      "35036/35036 [==============================] - 1s 32us/step - loss: 0.0891 - val_loss: 0.1310\n",
      "Epoch 43/200\n",
      "35036/35036 [==============================] - 1s 34us/step - loss: 0.0863 - val_loss: 0.1299\n",
      "Epoch 44/200\n",
      "35036/35036 [==============================] - 1s 36us/step - loss: 0.0908 - val_loss: 0.1145\n",
      "Epoch 45/200\n",
      "35036/35036 [==============================] - 1s 41us/step - loss: 0.0936 - val_loss: 0.1230\n",
      "Epoch 46/200\n",
      "35036/35036 [==============================] - 1s 41us/step - loss: 0.0856 - val_loss: 0.1375\n",
      "Epoch 47/200\n",
      "35036/35036 [==============================] - 1s 38us/step - loss: 0.0908 - val_loss: 0.1558\n",
      "Epoch 48/200\n",
      "35036/35036 [==============================] - 1s 42us/step - loss: 0.0906 - val_loss: 0.1190\n",
      "Epoch 49/200\n",
      "35036/35036 [==============================] - 2s 46us/step - loss: 0.0899 - val_loss: 0.1191\n",
      "Epoch 50/200\n",
      "35036/35036 [==============================] - 2s 48us/step - loss: 0.0911 - val_loss: 0.1384\n",
      "Epoch 51/200\n",
      "35036/35036 [==============================] - 2s 50us/step - loss: 0.0892 - val_loss: 0.1589\n",
      "Epoch 52/200\n",
      "35036/35036 [==============================] - 2s 52us/step - loss: 0.0894 - val_loss: 0.1643\n",
      "Epoch 53/200\n",
      "35036/35036 [==============================] - 2s 53us/step - loss: 0.0897 - val_loss: 0.1376\n",
      "Epoch 54/200\n",
      "35036/35036 [==============================] - 2s 51us/step - loss: 0.0928 - val_loss: 0.1403\n",
      "Epoch 55/200\n",
      "35036/35036 [==============================] - 2s 49us/step - loss: 0.0902 - val_loss: 0.1281\n",
      "Epoch 56/200\n",
      "35036/35036 [==============================] - 2s 51us/step - loss: 0.0877 - val_loss: 0.1381\n",
      "Epoch 57/200\n",
      "35036/35036 [==============================] - 2s 52us/step - loss: 0.0862 - val_loss: 0.1515\n",
      "Epoch 58/200\n",
      "35036/35036 [==============================] - 2s 51us/step - loss: 0.0883 - val_loss: 0.1437\n",
      "Epoch 59/200\n",
      "35036/35036 [==============================] - 2s 48us/step - loss: 0.0861 - val_loss: 0.0828\n",
      "Epoch 60/200\n",
      "35036/35036 [==============================] - 2s 48us/step - loss: 0.0942 - val_loss: 0.1626\n",
      "Epoch 61/200\n",
      "35036/35036 [==============================] - 2s 47us/step - loss: 0.0909 - val_loss: 0.2349\n",
      "Epoch 62/200\n",
      "35036/35036 [==============================] - 2s 46us/step - loss: 0.0928 - val_loss: 0.1133\n",
      "Epoch 63/200\n",
      "35036/35036 [==============================] - 2s 47us/step - loss: 0.0869 - val_loss: 0.1374\n",
      "Epoch 64/200\n",
      "35036/35036 [==============================] - 2s 49us/step - loss: 0.0933 - val_loss: 0.1461\n",
      "Epoch 65/200\n",
      "35036/35036 [==============================] - 2s 49us/step - loss: 0.0901 - val_loss: 0.1060\n",
      "Epoch 66/200\n",
      "35036/35036 [==============================] - 2s 48us/step - loss: 0.0928 - val_loss: 0.1041\n",
      "Epoch 67/200\n",
      "35036/35036 [==============================] - 2s 50us/step - loss: 0.0910 - val_loss: 0.1426\n",
      "Epoch 68/200\n",
      "35036/35036 [==============================] - 2s 50us/step - loss: 0.0962 - val_loss: 0.1427\n",
      "Epoch 69/200\n",
      "35036/35036 [==============================] - 2s 50us/step - loss: 0.0881 - val_loss: 0.1377\n",
      "Epoch 70/200\n",
      "35036/35036 [==============================] - 2s 49us/step - loss: 0.0901 - val_loss: 0.1152\n",
      "Epoch 71/200\n",
      "35036/35036 [==============================] - 2s 49us/step - loss: 0.0952 - val_loss: 0.1744\n",
      "Epoch 72/200\n",
      "35036/35036 [==============================] - 2s 47us/step - loss: 0.0930 - val_loss: 0.1268\n",
      "Epoch 73/200\n",
      "35036/35036 [==============================] - 2s 48us/step - loss: 0.0882 - val_loss: 0.1477\n",
      "Epoch 74/200\n",
      "35036/35036 [==============================] - 2s 52us/step - loss: 0.0871 - val_loss: 0.1971\n",
      "Epoch 75/200\n",
      "35036/35036 [==============================] - 2s 47us/step - loss: 0.0922 - val_loss: 0.1242\n",
      "Epoch 76/200\n",
      "35036/35036 [==============================] - 2s 51us/step - loss: 0.0904 - val_loss: 0.1452\n",
      "Epoch 77/200\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "35036/35036 [==============================] - 2s 52us/step - loss: 0.0914 - val_loss: 0.1704\n",
      "Epoch 78/200\n",
      "35036/35036 [==============================] - 2s 51us/step - loss: 0.0907 - val_loss: 0.1294\n",
      "Epoch 79/200\n",
      "35036/35036 [==============================] - 2s 51us/step - loss: 0.0920 - val_loss: 0.1451\n",
      "Epoch 80/200\n",
      "35036/35036 [==============================] - 2s 49us/step - loss: 0.0876 - val_loss: 0.1305\n",
      "Epoch 81/200\n",
      "35036/35036 [==============================] - 2s 50us/step - loss: 0.0923 - val_loss: 0.1545\n",
      "Epoch 82/200\n",
      "35036/35036 [==============================] - 2s 46us/step - loss: 0.0907 - val_loss: 0.2093\n",
      "Epoch 83/200\n",
      "35036/35036 [==============================] - 2s 49us/step - loss: 0.0911 - val_loss: 0.1125\n",
      "Epoch 84/200\n",
      "35036/35036 [==============================] - 2s 48us/step - loss: 0.0921 - val_loss: 0.1371\n",
      "Epoch 85/200\n",
      "35036/35036 [==============================] - 2s 52us/step - loss: 0.0921 - val_loss: 0.1223\n",
      "Epoch 86/200\n",
      "35036/35036 [==============================] - 2s 51us/step - loss: 0.0895 - val_loss: 0.1327\n",
      "Epoch 87/200\n",
      "35036/35036 [==============================] - 2s 51us/step - loss: 0.0922 - val_loss: 0.1342\n",
      "Epoch 88/200\n",
      "35036/35036 [==============================] - 2s 48us/step - loss: 0.0892 - val_loss: 0.1425\n",
      "Epoch 89/200\n",
      "35036/35036 [==============================] - 2s 48us/step - loss: 0.0905 - val_loss: 0.1401\n",
      "Epoch 90/200\n",
      "35036/35036 [==============================] - 2s 51us/step - loss: 0.0865 - val_loss: 0.1598\n",
      "Epoch 91/200\n",
      "35036/35036 [==============================] - 2s 53us/step - loss: 0.0921 - val_loss: 0.1482\n",
      "Epoch 92/200\n",
      "35036/35036 [==============================] - 2s 54us/step - loss: 0.0904 - val_loss: 0.1812\n",
      "Epoch 93/200\n",
      "35036/35036 [==============================] - 2s 56us/step - loss: 0.0890 - val_loss: 0.1538\n",
      "Epoch 94/200\n",
      "35036/35036 [==============================] - 2s 58us/step - loss: 0.0912 - val_loss: 0.1751\n",
      "Epoch 95/200\n",
      "35036/35036 [==============================] - 2s 60us/step - loss: 0.0935 - val_loss: 0.1446\n",
      "Epoch 96/200\n",
      "35036/35036 [==============================] - 2s 58us/step - loss: 0.0959 - val_loss: 0.1704\n",
      "Epoch 97/200\n",
      "35036/35036 [==============================] - 2s 56us/step - loss: 0.0898 - val_loss: 0.1340\n",
      "Epoch 98/200\n",
      "35036/35036 [==============================] - 2s 56us/step - loss: 0.0926 - val_loss: 0.1191\n",
      "Epoch 99/200\n",
      "35036/35036 [==============================] - 2s 55us/step - loss: 0.0922 - val_loss: 0.1858\n",
      "Epoch 100/200\n",
      "35036/35036 [==============================] - 2s 54us/step - loss: 0.0890 - val_loss: 0.1660\n",
      "Epoch 101/200\n",
      "35036/35036 [==============================] - 2s 55us/step - loss: 0.0891 - val_loss: 0.1529\n",
      "Epoch 102/200\n",
      "35036/35036 [==============================] - 2s 53us/step - loss: 0.0917 - val_loss: 0.1672\n",
      "Epoch 103/200\n",
      "35036/35036 [==============================] - 2s 50us/step - loss: 0.0896 - val_loss: 0.1984\n",
      "Epoch 104/200\n",
      "35036/35036 [==============================] - 2s 55us/step - loss: 0.0984 - val_loss: 0.1904\n",
      "Epoch 105/200\n",
      "35036/35036 [==============================] - 2s 56us/step - loss: 0.0927 - val_loss: 0.1470\n",
      "Epoch 106/200\n",
      "35036/35036 [==============================] - 2s 52us/step - loss: 0.0919 - val_loss: 0.1847\n",
      "Epoch 107/200\n",
      "35036/35036 [==============================] - 2s 61us/step - loss: 0.0944 - val_loss: 0.1778\n",
      "Epoch 108/200\n",
      "35036/35036 [==============================] - 2s 55us/step - loss: 0.0909 - val_loss: 0.2202\n",
      "Epoch 109/200\n",
      "35036/35036 [==============================] - 2s 56us/step - loss: 0.0904 - val_loss: 0.1544\n",
      "Epoch 110/200\n",
      "35036/35036 [==============================] - 2s 60us/step - loss: 0.0935 - val_loss: 0.2057\n",
      "Epoch 111/200\n",
      "35036/35036 [==============================] - 2s 60us/step - loss: 0.0906 - val_loss: 0.1687\n",
      "Epoch 112/200\n",
      "35036/35036 [==============================] - 2s 55us/step - loss: 0.0888 - val_loss: 0.1364\n",
      "Epoch 113/200\n",
      "35036/35036 [==============================] - 2s 55us/step - loss: 0.0904 - val_loss: 0.1710\n",
      "Epoch 114/200\n",
      "35036/35036 [==============================] - 2s 55us/step - loss: 0.0965 - val_loss: 0.1640\n",
      "Epoch 115/200\n",
      "35036/35036 [==============================] - 2s 51us/step - loss: 0.0910 - val_loss: 0.2066\n",
      "Epoch 116/200\n",
      "35036/35036 [==============================] - 2s 51us/step - loss: 0.0921 - val_loss: 0.2123\n",
      "Epoch 117/200\n",
      "35036/35036 [==============================] - 2s 50us/step - loss: 0.0893 - val_loss: 0.2597\n",
      "Epoch 118/200\n",
      "35036/35036 [==============================] - 2s 49us/step - loss: 0.0898 - val_loss: 0.2116\n",
      "Epoch 119/200\n",
      "35036/35036 [==============================] - 2s 49us/step - loss: 0.0902 - val_loss: 0.1684\n",
      "Epoch 120/200\n",
      "35036/35036 [==============================] - 2s 50us/step - loss: 0.0991 - val_loss: 0.2490\n",
      "Epoch 121/200\n",
      "35036/35036 [==============================] - 2s 52us/step - loss: 0.0958 - val_loss: 0.2495\n",
      "Epoch 122/200\n",
      "35036/35036 [==============================] - 2s 47us/step - loss: 0.0958 - val_loss: 0.2965\n",
      "Epoch 123/200\n",
      "35036/35036 [==============================] - 2s 48us/step - loss: 0.0944 - val_loss: 0.2083\n",
      "Epoch 124/200\n",
      "35036/35036 [==============================] - 2s 48us/step - loss: 0.0962 - val_loss: 0.2091\n",
      "Epoch 125/200\n",
      "35036/35036 [==============================] - 2s 46us/step - loss: 0.0924 - val_loss: 0.2076\n",
      "Epoch 126/200\n",
      "35036/35036 [==============================] - 2s 47us/step - loss: 0.0895 - val_loss: 0.2364\n",
      "Epoch 127/200\n",
      "35036/35036 [==============================] - 2s 46us/step - loss: 0.0895 - val_loss: 0.2862\n",
      "Epoch 128/200\n",
      "35036/35036 [==============================] - 2s 47us/step - loss: 0.0938 - val_loss: 0.2277\n",
      "Epoch 129/200\n",
      "35036/35036 [==============================] - 2s 47us/step - loss: 0.0899 - val_loss: 0.2039\n",
      "Epoch 130/200\n",
      "35036/35036 [==============================] - 2s 45us/step - loss: 0.0925 - val_loss: 0.2476\n",
      "Epoch 131/200\n",
      "35036/35036 [==============================] - 2s 46us/step - loss: 0.0860 - val_loss: 0.2593\n",
      "Epoch 132/200\n",
      "35036/35036 [==============================] - 2s 45us/step - loss: 0.0873 - val_loss: 0.2521\n",
      "Epoch 133/200\n",
      "35036/35036 [==============================] - 2s 48us/step - loss: 0.0967 - val_loss: 0.2098\n",
      "Epoch 134/200\n",
      "35036/35036 [==============================] - 2s 48us/step - loss: 0.0924 - val_loss: 0.2341\n",
      "Epoch 135/200\n",
      "35036/35036 [==============================] - 2s 49us/step - loss: 0.0872 - val_loss: 0.2323\n",
      "Epoch 136/200\n",
      "35036/35036 [==============================] - 2s 49us/step - loss: 0.0884 - val_loss: 0.2002\n",
      "Epoch 137/200\n",
      "35036/35036 [==============================] - 2s 55us/step - loss: 0.0995 - val_loss: 0.2223\n",
      "Epoch 138/200\n",
      "35036/35036 [==============================] - 2s 52us/step - loss: 0.0914 - val_loss: 0.2359\n",
      "Epoch 139/200\n",
      "35036/35036 [==============================] - 2s 52us/step - loss: 0.0917 - val_loss: 0.2202\n",
      "Epoch 140/200\n",
      "35036/35036 [==============================] - 2s 53us/step - loss: 0.0894 - val_loss: 0.2040\n",
      "Epoch 141/200\n",
      "35036/35036 [==============================] - 2s 53us/step - loss: 0.0919 - val_loss: 0.1880\n",
      "Epoch 142/200\n",
      "35036/35036 [==============================] - 2s 53us/step - loss: 0.0882 - val_loss: 0.2005\n",
      "Epoch 143/200\n",
      "35036/35036 [==============================] - 2s 56us/step - loss: 0.0968 - val_loss: 0.2259\n",
      "Epoch 144/200\n",
      "35036/35036 [==============================] - 2s 57us/step - loss: 0.0964 - val_loss: 0.2468\n",
      "Epoch 145/200\n",
      "35036/35036 [==============================] - 2s 55us/step - loss: 0.0929 - val_loss: 0.2468\n",
      "Epoch 146/200\n",
      "35036/35036 [==============================] - 2s 52us/step - loss: 0.0902 - val_loss: 0.1986\n",
      "Epoch 147/200\n",
      "35036/35036 [==============================] - 2s 53us/step - loss: 0.1030 - val_loss: 0.2633\n",
      "Epoch 148/200\n",
      "35036/35036 [==============================] - 2s 55us/step - loss: 0.1007 - val_loss: 0.2049\n",
      "Epoch 149/200\n",
      "35036/35036 [==============================] - 2s 53us/step - loss: 0.0856 - val_loss: 0.2634\n",
      "Epoch 150/200\n",
      "35036/35036 [==============================] - 2s 59us/step - loss: 0.0899 - val_loss: 0.2575\n",
      "Epoch 151/200\n",
      "35036/35036 [==============================] - 2s 56us/step - loss: 0.0886 - val_loss: 0.2843\n",
      "Epoch 152/200\n",
      "35036/35036 [==============================] - 2s 51us/step - loss: 0.0977 - val_loss: 0.2449\n",
      "Epoch 153/200\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "35036/35036 [==============================] - 2s 51us/step - loss: 0.0956 - val_loss: 0.2356\n",
      "Epoch 154/200\n",
      "35036/35036 [==============================] - 2s 53us/step - loss: 0.0963 - val_loss: 0.2160\n",
      "Epoch 155/200\n",
      "35036/35036 [==============================] - 2s 59us/step - loss: 0.0961 - val_loss: 0.2483\n",
      "Epoch 156/200\n",
      "35036/35036 [==============================] - 2s 49us/step - loss: 0.1036 - val_loss: 0.2023\n",
      "Epoch 157/200\n",
      "35036/35036 [==============================] - 2s 49us/step - loss: 0.0872 - val_loss: 0.2920\n",
      "Epoch 158/200\n",
      "35036/35036 [==============================] - 2s 52us/step - loss: 0.0900 - val_loss: 0.3016\n",
      "Epoch 159/200\n",
      "35036/35036 [==============================] - 2s 52us/step - loss: 0.0909 - val_loss: 0.1626\n",
      "Epoch 160/200\n",
      "35036/35036 [==============================] - 2s 51us/step - loss: 0.0949 - val_loss: 0.2435\n",
      "Epoch 161/200\n",
      "35036/35036 [==============================] - 2s 50us/step - loss: 0.0959 - val_loss: 0.2456\n",
      "Epoch 162/200\n",
      "35036/35036 [==============================] - 2s 50us/step - loss: 0.0910 - val_loss: 0.2565\n",
      "Epoch 163/200\n",
      "35036/35036 [==============================] - 2s 48us/step - loss: 0.0883 - val_loss: 0.3149\n",
      "Epoch 164/200\n",
      "35036/35036 [==============================] - 2s 53us/step - loss: 0.0920 - val_loss: 0.1686\n",
      "Epoch 165/200\n",
      "35036/35036 [==============================] - 2s 48us/step - loss: 0.0905 - val_loss: 0.2144\n",
      "Epoch 166/200\n",
      "35036/35036 [==============================] - 2s 54us/step - loss: 0.1043 - val_loss: 0.2282\n",
      "Epoch 167/200\n",
      "35036/35036 [==============================] - 2s 53us/step - loss: 0.0913 - val_loss: 0.1890\n",
      "Epoch 168/200\n",
      "35036/35036 [==============================] - 2s 49us/step - loss: 0.0948 - val_loss: 0.2110\n",
      "Epoch 169/200\n",
      "35036/35036 [==============================] - 2s 56us/step - loss: 0.0892 - val_loss: 0.2458\n",
      "Epoch 170/200\n",
      "35036/35036 [==============================] - 2s 57us/step - loss: 0.0916 - val_loss: 0.1809\n",
      "Epoch 171/200\n",
      "35036/35036 [==============================] - 2s 56us/step - loss: 0.0911 - val_loss: 0.2242\n",
      "Epoch 172/200\n",
      "35036/35036 [==============================] - 2s 53us/step - loss: 0.1016 - val_loss: 0.2263\n",
      "Epoch 173/200\n",
      "35036/35036 [==============================] - 2s 54us/step - loss: 0.0911 - val_loss: 0.2345\n",
      "Epoch 174/200\n",
      "35036/35036 [==============================] - 2s 53us/step - loss: 0.0941 - val_loss: 0.1729\n",
      "Epoch 175/200\n",
      "35036/35036 [==============================] - 2s 51us/step - loss: 0.0981 - val_loss: 0.2438\n",
      "Epoch 176/200\n",
      "35036/35036 [==============================] - 2s 52us/step - loss: 0.0871 - val_loss: 0.2057\n",
      "Epoch 177/200\n",
      "35036/35036 [==============================] - 2s 54us/step - loss: 0.0951 - val_loss: 0.2226\n",
      "Epoch 178/200\n",
      "35036/35036 [==============================] - 2s 55us/step - loss: 0.1026 - val_loss: 0.2434\n",
      "Epoch 179/200\n",
      "35036/35036 [==============================] - 2s 55us/step - loss: 0.0945 - val_loss: 0.1752\n",
      "Epoch 180/200\n",
      "35036/35036 [==============================] - 2s 54us/step - loss: 0.0948 - val_loss: 0.2693\n",
      "Epoch 181/200\n",
      "35036/35036 [==============================] - 2s 53us/step - loss: 0.0978 - val_loss: 0.2246\n",
      "Epoch 182/200\n",
      "35036/35036 [==============================] - 2s 53us/step - loss: 0.1038 - val_loss: 0.2192\n",
      "Epoch 183/200\n",
      "35036/35036 [==============================] - 2s 52us/step - loss: 0.0949 - val_loss: 0.2024\n",
      "Epoch 184/200\n",
      "35036/35036 [==============================] - 2s 52us/step - loss: 0.0911 - val_loss: 0.2192\n",
      "Epoch 185/200\n",
      "35036/35036 [==============================] - 2s 52us/step - loss: 0.0911 - val_loss: 0.1809\n",
      "Epoch 186/200\n",
      "35036/35036 [==============================] - 2s 54us/step - loss: 0.1007 - val_loss: 0.1979\n",
      "Epoch 187/200\n",
      "35036/35036 [==============================] - 2s 53us/step - loss: 0.0969 - val_loss: 0.3035\n",
      "Epoch 188/200\n",
      "35036/35036 [==============================] - 2s 55us/step - loss: 0.0947 - val_loss: 0.2531\n",
      "Epoch 189/200\n",
      "35036/35036 [==============================] - 2s 58us/step - loss: 0.0952 - val_loss: 0.2377\n",
      "Epoch 190/200\n",
      "35036/35036 [==============================] - 2s 54us/step - loss: 0.0920 - val_loss: 0.2309\n",
      "Epoch 191/200\n",
      "35036/35036 [==============================] - 2s 55us/step - loss: 0.0913 - val_loss: 0.2561\n",
      "Epoch 192/200\n",
      "35036/35036 [==============================] - 2s 53us/step - loss: 0.0912 - val_loss: 0.2616\n",
      "Epoch 193/200\n",
      "35036/35036 [==============================] - 2s 51us/step - loss: 0.0912 - val_loss: 0.1913\n",
      "Epoch 194/200\n",
      "35036/35036 [==============================] - 2s 54us/step - loss: 0.0898 - val_loss: 0.2049\n",
      "Epoch 195/200\n",
      "35036/35036 [==============================] - 2s 55us/step - loss: 0.0968 - val_loss: 0.1889\n",
      "Epoch 196/200\n",
      "35036/35036 [==============================] - 2s 59us/step - loss: 0.0907 - val_loss: 0.1887\n",
      "Epoch 197/200\n",
      "35036/35036 [==============================] - 2s 58us/step - loss: 0.1015 - val_loss: 0.1979\n",
      "Epoch 198/200\n",
      "35036/35036 [==============================] - 2s 67us/step - loss: 0.0929 - val_loss: 0.1953\n",
      "Epoch 199/200\n",
      "35036/35036 [==============================] - 3s 89us/step - loss: 0.0901 - val_loss: 0.2095\n",
      "Epoch 200/200\n",
      "35036/35036 [==============================] - 3s 74us/step - loss: 0.1025 - val_loss: 0.2742\n"
     ]
    }
   ],
   "source": [
    "model.compile(loss='mean_squared_error', optimizer=Adam(lr=learning_rate))\n",
    "history = model.fit(X_train_sc,y_train_sc,\n",
    "                    batch_size=batch,\n",
    "                    epochs=training_epochs,\n",
    "                    validation_data=(X_test_sc, y_test_sc)) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x1a2872cc50>]"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(history.history['loss'])\n",
    "plt.plot(history.history['val_loss'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x1a28cf2518>"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "y_pred = model.predict(X_test_sc)\n",
    "plt.plot(y_test_sc, label='True')\n",
    "plt.plot(y_pred, label='Predicted')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Time Series LSTM"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_tr_t = X_train_sc.reshape(X_train_sc.shape[0], 1, X_train_sc.shape[1])\n",
    "X_tst_t = X_test_sc.reshape(X_test_sc.shape[0], 1, X_test_sc.shape[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 35036 samples, validate on 8741 samples\n",
      "Epoch 1/200\n",
      "35036/35036 [==============================] - 5s 137us/step - loss: 0.2120 - val_loss: 0.0262\n",
      "Epoch 2/200\n",
      "35036/35036 [==============================] - 2s 70us/step - loss: 0.0229 - val_loss: 0.0249\n",
      "Epoch 3/200\n",
      "35036/35036 [==============================] - 3s 78us/step - loss: 0.0210 - val_loss: 0.0231\n",
      "Epoch 4/200\n",
      "35036/35036 [==============================] - 2s 70us/step - loss: 0.0188 - val_loss: 0.0174\n",
      "Epoch 5/200\n",
      "35036/35036 [==============================] - 3s 72us/step - loss: 0.0169 - val_loss: 0.0165\n",
      "Epoch 6/200\n",
      "35036/35036 [==============================] - 2s 69us/step - loss: 0.0169 - val_loss: 0.0160\n",
      "Epoch 7/200\n",
      "35036/35036 [==============================] - 2s 69us/step - loss: 0.0165 - val_loss: 0.0171\n",
      "Epoch 8/200\n",
      "35036/35036 [==============================] - 2s 67us/step - loss: 0.0154 - val_loss: 0.0166\n",
      "Epoch 9/200\n",
      "35036/35036 [==============================] - 2s 67us/step - loss: 0.0162 - val_loss: 0.0164\n",
      "Epoch 10/200\n",
      "35036/35036 [==============================] - 2s 67us/step - loss: 0.0154 - val_loss: 0.0157\n",
      "Epoch 11/200\n",
      "35036/35036 [==============================] - 2s 69us/step - loss: 0.0152 - val_loss: 0.0183\n",
      "Epoch 12/200\n",
      "35036/35036 [==============================] - 2s 69us/step - loss: 0.0164 - val_loss: 0.0200\n",
      "Epoch 13/200\n",
      "35036/35036 [==============================] - 2s 65us/step - loss: 0.0157 - val_loss: 0.0151\n",
      "Epoch 14/200\n",
      "35036/35036 [==============================] - 2s 68us/step - loss: 0.0144 - val_loss: 0.0153\n",
      "Epoch 15/200\n",
      "35036/35036 [==============================] - 2s 66us/step - loss: 0.0157 - val_loss: 0.0149\n",
      "Epoch 16/200\n",
      "35036/35036 [==============================] - 3s 75us/step - loss: 0.0159 - val_loss: 0.0147\n",
      "Epoch 17/200\n",
      "35036/35036 [==============================] - 2s 70us/step - loss: 0.0144 - val_loss: 0.0143\n",
      "Epoch 18/200\n",
      "35036/35036 [==============================] - 2s 68us/step - loss: 0.0147 - val_loss: 0.0201\n",
      "Epoch 19/200\n",
      "35036/35036 [==============================] - 2s 67us/step - loss: 0.0149 - val_loss: 0.0144\n",
      "Epoch 20/200\n",
      "35036/35036 [==============================] - 2s 67us/step - loss: 0.0146 - val_loss: 0.0145\n",
      "Epoch 21/200\n",
      "35036/35036 [==============================] - 2s 69us/step - loss: 0.0162 - val_loss: 0.0159\n",
      "Epoch 22/200\n",
      "35036/35036 [==============================] - 2s 70us/step - loss: 0.0144 - val_loss: 0.0151\n",
      "Epoch 23/200\n",
      "35036/35036 [==============================] - 2s 69us/step - loss: 0.0143 - val_loss: 0.0163\n",
      "Epoch 24/200\n",
      "35036/35036 [==============================] - 2s 69us/step - loss: 0.0145 - val_loss: 0.0146\n",
      "Epoch 25/200\n",
      "35036/35036 [==============================] - 3s 71us/step - loss: 0.0151 - val_loss: 0.0218\n",
      "Epoch 26/200\n",
      "35036/35036 [==============================] - 2s 70us/step - loss: 0.0149 - val_loss: 0.0140\n",
      "Epoch 27/200\n",
      "35036/35036 [==============================] - 3s 73us/step - loss: 0.0133 - val_loss: 0.0178\n",
      "Epoch 28/200\n",
      "35036/35036 [==============================] - 3s 78us/step - loss: 0.0156 - val_loss: 0.0178\n",
      "Epoch 29/200\n",
      "35036/35036 [==============================] - 2s 70us/step - loss: 0.0144 - val_loss: 0.0149\n",
      "Epoch 30/200\n",
      "35036/35036 [==============================] - 2s 63us/step - loss: 0.0138 - val_loss: 0.0139\n",
      "Epoch 31/200\n",
      "35036/35036 [==============================] - 2s 67us/step - loss: 0.0136 - val_loss: 0.0154\n",
      "Epoch 32/200\n",
      "35036/35036 [==============================] - 2s 66us/step - loss: 0.0143 - val_loss: 0.0151\n",
      "Epoch 33/200\n",
      "35036/35036 [==============================] - 2s 68us/step - loss: 0.0137 - val_loss: 0.0164\n",
      "Epoch 34/200\n",
      "35036/35036 [==============================] - 2s 68us/step - loss: 0.0149 - val_loss: 0.0145\n",
      "Epoch 35/200\n",
      "35036/35036 [==============================] - 2s 69us/step - loss: 0.0142 - val_loss: 0.0234\n",
      "Epoch 36/200\n",
      "35036/35036 [==============================] - 2s 68us/step - loss: 0.0139 - val_loss: 0.0147\n",
      "Epoch 37/200\n",
      "35036/35036 [==============================] - 3s 73us/step - loss: 0.0144 - val_loss: 0.0158\n",
      "Epoch 38/200\n",
      "35036/35036 [==============================] - 2s 69us/step - loss: 0.0138 - val_loss: 0.0152\n",
      "Epoch 39/200\n",
      "35036/35036 [==============================] - 2s 71us/step - loss: 0.0138 - val_loss: 0.0145\n",
      "Epoch 40/200\n",
      "35036/35036 [==============================] - 3s 72us/step - loss: 0.0144 - val_loss: 0.0159\n",
      "Epoch 41/200\n",
      "35036/35036 [==============================] - 2s 71us/step - loss: 0.0140 - val_loss: 0.0177\n",
      "Epoch 42/200\n",
      "35036/35036 [==============================] - 2s 67us/step - loss: 0.0138 - val_loss: 0.0140\n",
      "Epoch 43/200\n",
      "35036/35036 [==============================] - 2s 64us/step - loss: 0.0139 - val_loss: 0.0143\n",
      "Epoch 44/200\n",
      "35036/35036 [==============================] - 2s 68us/step - loss: 0.0134 - val_loss: 0.0177\n",
      "Epoch 45/200\n",
      "35036/35036 [==============================] - 2s 68us/step - loss: 0.0134 - val_loss: 0.0146\n",
      "Epoch 46/200\n",
      "35036/35036 [==============================] - 2s 65us/step - loss: 0.0143 - val_loss: 0.0204\n",
      "Epoch 47/200\n",
      "35036/35036 [==============================] - 2s 65us/step - loss: 0.0145 - val_loss: 0.0170\n",
      "Epoch 48/200\n",
      "35036/35036 [==============================] - 2s 64us/step - loss: 0.0131 - val_loss: 0.0136\n",
      "Epoch 49/200\n",
      "35036/35036 [==============================] - 2s 67us/step - loss: 0.0134 - val_loss: 0.0146\n",
      "Epoch 50/200\n",
      "35036/35036 [==============================] - 2s 66us/step - loss: 0.0134 - val_loss: 0.0139\n",
      "Epoch 51/200\n",
      "35036/35036 [==============================] - 2s 65us/step - loss: 0.0141 - val_loss: 0.0160\n",
      "Epoch 52/200\n",
      "35036/35036 [==============================] - 2s 67us/step - loss: 0.0152 - val_loss: 0.0156\n",
      "Epoch 53/200\n",
      "35036/35036 [==============================] - 2s 66us/step - loss: 0.0136 - val_loss: 0.0150\n",
      "Epoch 54/200\n",
      "35036/35036 [==============================] - 2s 65us/step - loss: 0.0141 - val_loss: 0.0148\n",
      "Epoch 55/200\n",
      "35036/35036 [==============================] - 2s 66us/step - loss: 0.0128 - val_loss: 0.0137\n",
      "Epoch 56/200\n",
      "35036/35036 [==============================] - 2s 67us/step - loss: 0.0135 - val_loss: 0.0137\n",
      "Epoch 57/200\n",
      "35036/35036 [==============================] - 2s 68us/step - loss: 0.0134 - val_loss: 0.0142\n",
      "Epoch 58/200\n",
      "35036/35036 [==============================] - 2s 70us/step - loss: 0.0140 - val_loss: 0.0145\n",
      "Epoch 59/200\n",
      "35036/35036 [==============================] - 2s 70us/step - loss: 0.0134 - val_loss: 0.0141\n",
      "Epoch 60/200\n",
      "35036/35036 [==============================] - 2s 68us/step - loss: 0.0134 - val_loss: 0.0157\n",
      "Epoch 61/200\n",
      "35036/35036 [==============================] - 2s 69us/step - loss: 0.0150 - val_loss: 0.0189\n",
      "Epoch 62/200\n",
      "35036/35036 [==============================] - 2s 71us/step - loss: 0.0130 - val_loss: 0.0140\n",
      "Epoch 63/200\n",
      "35036/35036 [==============================] - 2s 69us/step - loss: 0.0135 - val_loss: 0.0135\n",
      "Epoch 64/200\n",
      "35036/35036 [==============================] - 2s 69us/step - loss: 0.0144 - val_loss: 0.0179\n",
      "Epoch 65/200\n",
      "35036/35036 [==============================] - 2s 71us/step - loss: 0.0133 - val_loss: 0.0167\n",
      "Epoch 66/200\n",
      "35036/35036 [==============================] - 2s 71us/step - loss: 0.0132 - val_loss: 0.0158\n",
      "Epoch 67/200\n",
      "35036/35036 [==============================] - 2s 68us/step - loss: 0.0134 - val_loss: 0.0134\n",
      "Epoch 68/200\n",
      "35036/35036 [==============================] - 2s 67us/step - loss: 0.0133 - val_loss: 0.0151\n",
      "Epoch 69/200\n",
      "35036/35036 [==============================] - 2s 69us/step - loss: 0.0132 - val_loss: 0.0146\n",
      "Epoch 70/200\n",
      "35036/35036 [==============================] - 2s 68us/step - loss: 0.0130 - val_loss: 0.0133\n",
      "Epoch 71/200\n",
      "35036/35036 [==============================] - 3s 71us/step - loss: 0.0130 - val_loss: 0.0160\n",
      "Epoch 72/200\n",
      "35036/35036 [==============================] - 2s 70us/step - loss: 0.0139 - val_loss: 0.0154\n",
      "Epoch 73/200\n",
      "35036/35036 [==============================] - 2s 70us/step - loss: 0.0134 - val_loss: 0.0135\n",
      "Epoch 74/200\n",
      "35036/35036 [==============================] - 2s 69us/step - loss: 0.0137 - val_loss: 0.0143\n",
      "Epoch 75/200\n",
      "35036/35036 [==============================] - 3s 71us/step - loss: 0.0129 - val_loss: 0.0139\n",
      "Epoch 76/200\n",
      "35036/35036 [==============================] - 2s 69us/step - loss: 0.0128 - val_loss: 0.0161\n",
      "Epoch 77/200\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "35036/35036 [==============================] - 2s 66us/step - loss: 0.0137 - val_loss: 0.0172\n",
      "Epoch 78/200\n",
      "35036/35036 [==============================] - 2s 65us/step - loss: 0.0130 - val_loss: 0.0168\n",
      "Epoch 79/200\n",
      "35036/35036 [==============================] - 2s 66us/step - loss: 0.0139 - val_loss: 0.0147\n",
      "Epoch 80/200\n",
      "35036/35036 [==============================] - 2s 65us/step - loss: 0.0125 - val_loss: 0.0144\n",
      "Epoch 81/200\n",
      "35036/35036 [==============================] - 2s 66us/step - loss: 0.0132 - val_loss: 0.0151\n",
      "Epoch 82/200\n",
      "35036/35036 [==============================] - 2s 66us/step - loss: 0.0135 - val_loss: 0.0146\n",
      "Epoch 83/200\n",
      "35036/35036 [==============================] - 2s 69us/step - loss: 0.0140 - val_loss: 0.0137\n",
      "Epoch 84/200\n",
      "35036/35036 [==============================] - 2s 64us/step - loss: 0.0127 - val_loss: 0.0157\n",
      "Epoch 85/200\n",
      "35036/35036 [==============================] - 2s 63us/step - loss: 0.0137 - val_loss: 0.0149\n",
      "Epoch 86/200\n",
      "35036/35036 [==============================] - 2s 64us/step - loss: 0.0133 - val_loss: 0.0144\n",
      "Epoch 87/200\n",
      "35036/35036 [==============================] - 2s 67us/step - loss: 0.0130 - val_loss: 0.0147\n",
      "Epoch 88/200\n",
      "35036/35036 [==============================] - 2s 63us/step - loss: 0.0129 - val_loss: 0.0221\n",
      "Epoch 89/200\n",
      "35036/35036 [==============================] - 2s 63us/step - loss: 0.0132 - val_loss: 0.0147\n",
      "Epoch 90/200\n",
      "35036/35036 [==============================] - 2s 62us/step - loss: 0.0132 - val_loss: 0.0162\n",
      "Epoch 91/200\n",
      "35036/35036 [==============================] - 2s 65us/step - loss: 0.0135 - val_loss: 0.0140\n",
      "Epoch 92/200\n",
      "35036/35036 [==============================] - 2s 64us/step - loss: 0.0129 - val_loss: 0.0142\n",
      "Epoch 93/200\n",
      "35036/35036 [==============================] - 2s 62us/step - loss: 0.0136 - val_loss: 0.0154\n",
      "Epoch 94/200\n",
      "35036/35036 [==============================] - 2s 64us/step - loss: 0.0129 - val_loss: 0.0138\n",
      "Epoch 95/200\n",
      "35036/35036 [==============================] - 2s 62us/step - loss: 0.0132 - val_loss: 0.0135\n",
      "Epoch 96/200\n",
      "35036/35036 [==============================] - 2s 67us/step - loss: 0.0129 - val_loss: 0.0141\n",
      "Epoch 97/200\n",
      "35036/35036 [==============================] - 2s 66us/step - loss: 0.0132 - val_loss: 0.0157\n",
      "Epoch 98/200\n",
      "35036/35036 [==============================] - 2s 66us/step - loss: 0.0131 - val_loss: 0.0142\n",
      "Epoch 99/200\n",
      "35036/35036 [==============================] - 2s 66us/step - loss: 0.0132 - val_loss: 0.0154\n",
      "Epoch 100/200\n",
      "35036/35036 [==============================] - 2s 65us/step - loss: 0.0135 - val_loss: 0.0153\n",
      "Epoch 101/200\n",
      "35036/35036 [==============================] - 2s 67us/step - loss: 0.0131 - val_loss: 0.0134\n",
      "Epoch 102/200\n",
      "35036/35036 [==============================] - 3s 72us/step - loss: 0.0130 - val_loss: 0.0201\n",
      "Epoch 103/200\n",
      "35036/35036 [==============================] - 2s 71us/step - loss: 0.0148 - val_loss: 0.0159\n",
      "Epoch 104/200\n",
      "35036/35036 [==============================] - 2s 71us/step - loss: 0.0142 - val_loss: 0.0142\n",
      "Epoch 105/200\n",
      "35036/35036 [==============================] - 2s 70us/step - loss: 0.0135 - val_loss: 0.0173\n",
      "Epoch 106/200\n",
      "35036/35036 [==============================] - 3s 72us/step - loss: 0.0130 - val_loss: 0.0141\n",
      "Epoch 107/200\n",
      "35036/35036 [==============================] - 2s 68us/step - loss: 0.0135 - val_loss: 0.0148\n",
      "Epoch 108/200\n",
      "35036/35036 [==============================] - 2s 68us/step - loss: 0.0144 - val_loss: 0.0173\n",
      "Epoch 109/200\n",
      "35036/35036 [==============================] - 2s 68us/step - loss: 0.0137 - val_loss: 0.0146\n",
      "Epoch 110/200\n",
      "35036/35036 [==============================] - 2s 68us/step - loss: 0.0128 - val_loss: 0.0137\n",
      "Epoch 111/200\n",
      "35036/35036 [==============================] - 2s 71us/step - loss: 0.0137 - val_loss: 0.0137\n",
      "Epoch 112/200\n",
      "35036/35036 [==============================] - 2s 69us/step - loss: 0.0135 - val_loss: 0.0161\n",
      "Epoch 113/200\n",
      "35036/35036 [==============================] - 2s 69us/step - loss: 0.0133 - val_loss: 0.0143\n",
      "Epoch 114/200\n",
      "35036/35036 [==============================] - 2s 71us/step - loss: 0.0132 - val_loss: 0.0142\n",
      "Epoch 115/200\n",
      "35036/35036 [==============================] - 3s 71us/step - loss: 0.0135 - val_loss: 0.0136\n",
      "Epoch 116/200\n",
      "35036/35036 [==============================] - 2s 70us/step - loss: 0.0137 - val_loss: 0.0138\n",
      "Epoch 117/200\n",
      "35036/35036 [==============================] - 2s 69us/step - loss: 0.0129 - val_loss: 0.0148\n",
      "Epoch 118/200\n",
      "35036/35036 [==============================] - 2s 66us/step - loss: 0.0132 - val_loss: 0.0151\n",
      "Epoch 119/200\n",
      "35036/35036 [==============================] - 2s 71us/step - loss: 0.0132 - val_loss: 0.0139\n",
      "Epoch 120/200\n",
      "35036/35036 [==============================] - 3s 76us/step - loss: 0.0133 - val_loss: 0.0151\n",
      "Epoch 121/200\n",
      "35036/35036 [==============================] - 3s 84us/step - loss: 0.0132 - val_loss: 0.0134\n",
      "Epoch 122/200\n",
      "35036/35036 [==============================] - 3s 77us/step - loss: 0.0133 - val_loss: 0.0182\n",
      "Epoch 123/200\n",
      "35036/35036 [==============================] - 2s 55us/step - loss: 0.0134 - val_loss: 0.0142\n",
      "Epoch 124/200\n",
      "35036/35036 [==============================] - 2s 54us/step - loss: 0.0129 - val_loss: 0.0143\n",
      "Epoch 125/200\n",
      "35036/35036 [==============================] - 2s 50us/step - loss: 0.0137 - val_loss: 0.0149\n",
      "Epoch 126/200\n",
      "35036/35036 [==============================] - 2s 49us/step - loss: 0.0132 - val_loss: 0.0139\n",
      "Epoch 127/200\n",
      "35036/35036 [==============================] - 2s 46us/step - loss: 0.0133 - val_loss: 0.0145\n",
      "Epoch 128/200\n",
      "35036/35036 [==============================] - 2s 43us/step - loss: 0.0137 - val_loss: 0.0139\n",
      "Epoch 129/200\n",
      "35036/35036 [==============================] - 2s 45us/step - loss: 0.0130 - val_loss: 0.0139\n",
      "Epoch 130/200\n",
      "35036/35036 [==============================] - 2s 45us/step - loss: 0.0133 - val_loss: 0.0139\n",
      "Epoch 131/200\n",
      "35036/35036 [==============================] - 2s 47us/step - loss: 0.0137 - val_loss: 0.0143\n",
      "Epoch 132/200\n",
      "35036/35036 [==============================] - 2s 49us/step - loss: 0.0128 - val_loss: 0.0138\n",
      "Epoch 133/200\n",
      "35036/35036 [==============================] - 2s 49us/step - loss: 0.0132 - val_loss: 0.0138\n",
      "Epoch 134/200\n",
      "35036/35036 [==============================] - 2s 50us/step - loss: 0.0134 - val_loss: 0.0141\n",
      "Epoch 135/200\n",
      "35036/35036 [==============================] - 2s 49us/step - loss: 0.0129 - val_loss: 0.0147\n",
      "Epoch 136/200\n",
      "35036/35036 [==============================] - 2s 56us/step - loss: 0.0132 - val_loss: 0.0140\n",
      "Epoch 137/200\n",
      "35036/35036 [==============================] - 2s 49us/step - loss: 0.0139 - val_loss: 0.0144\n",
      "Epoch 138/200\n",
      "35036/35036 [==============================] - 2s 50us/step - loss: 0.0132 - val_loss: 0.0143\n",
      "Epoch 139/200\n",
      "35036/35036 [==============================] - 1s 42us/step - loss: 0.0137 - val_loss: 0.0211\n",
      "Epoch 140/200\n",
      "35036/35036 [==============================] - 1s 41us/step - loss: 0.0139 - val_loss: 0.0156\n",
      "Epoch 141/200\n",
      "35036/35036 [==============================] - 2s 44us/step - loss: 0.0132 - val_loss: 0.0137\n",
      "Epoch 142/200\n",
      "35036/35036 [==============================] - 2s 47us/step - loss: 0.0133 - val_loss: 0.0138\n",
      "Epoch 143/200\n",
      "35036/35036 [==============================] - 2s 46us/step - loss: 0.0134 - val_loss: 0.0165\n",
      "Epoch 144/200\n",
      "35036/35036 [==============================] - 1s 42us/step - loss: 0.0143 - val_loss: 0.0152\n",
      "Epoch 145/200\n",
      "35036/35036 [==============================] - 1s 38us/step - loss: 0.0129 - val_loss: 0.0177\n",
      "Epoch 146/200\n",
      "35036/35036 [==============================] - 1s 37us/step - loss: 0.0126 - val_loss: 0.0135\n",
      "Epoch 147/200\n",
      "35036/35036 [==============================] - 1s 37us/step - loss: 0.0133 - val_loss: 0.0145\n",
      "Epoch 148/200\n",
      "35036/35036 [==============================] - 3s 73us/step - loss: 0.0133 - val_loss: 0.0134\n",
      "Epoch 149/200\n",
      "35036/35036 [==============================] - 1s 40us/step - loss: 0.0129 - val_loss: 0.0138\n",
      "Epoch 150/200\n",
      "35036/35036 [==============================] - 2s 48us/step - loss: 0.0131 - val_loss: 0.0138\n",
      "Epoch 151/200\n",
      "35036/35036 [==============================] - 2s 47us/step - loss: 0.0132 - val_loss: 0.0142\n",
      "Epoch 152/200\n",
      "35036/35036 [==============================] - 2s 51us/step - loss: 0.0130 - val_loss: 0.0151\n",
      "Epoch 153/200\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "35036/35036 [==============================] - 2s 57us/step - loss: 0.0129 - val_loss: 0.0238\n",
      "Epoch 154/200\n",
      "35036/35036 [==============================] - 2s 53us/step - loss: 0.0141 - val_loss: 0.0180\n",
      "Epoch 155/200\n",
      "35036/35036 [==============================] - 2s 52us/step - loss: 0.0130 - val_loss: 0.0159\n",
      "Epoch 156/200\n",
      "35036/35036 [==============================] - 2s 54us/step - loss: 0.0140 - val_loss: 0.0199\n",
      "Epoch 157/200\n",
      "35036/35036 [==============================] - 2s 53us/step - loss: 0.0136 - val_loss: 0.0149\n",
      "Epoch 158/200\n",
      "35036/35036 [==============================] - 2s 61us/step - loss: 0.0128 - val_loss: 0.0145\n",
      "Epoch 159/200\n",
      "35036/35036 [==============================] - 2s 51us/step - loss: 0.0130 - val_loss: 0.0144\n",
      "Epoch 160/200\n",
      "35036/35036 [==============================] - 2s 52us/step - loss: 0.0133 - val_loss: 0.0136\n",
      "Epoch 161/200\n",
      "35036/35036 [==============================] - 2s 48us/step - loss: 0.0137 - val_loss: 0.0173\n",
      "Epoch 162/200\n",
      "35036/35036 [==============================] - 2s 46us/step - loss: 0.0141 - val_loss: 0.0146\n",
      "Epoch 163/200\n",
      "35036/35036 [==============================] - 2s 47us/step - loss: 0.0132 - val_loss: 0.0224\n",
      "Epoch 164/200\n",
      "35036/35036 [==============================] - 2s 46us/step - loss: 0.0132 - val_loss: 0.0169\n",
      "Epoch 165/200\n",
      "35036/35036 [==============================] - 2s 47us/step - loss: 0.0131 - val_loss: 0.0137\n",
      "Epoch 166/200\n",
      "35036/35036 [==============================] - 2s 52us/step - loss: 0.0130 - val_loss: 0.0147\n",
      "Epoch 167/200\n",
      "35036/35036 [==============================] - 2s 49us/step - loss: 0.0137 - val_loss: 0.0156\n",
      "Epoch 168/200\n",
      "35036/35036 [==============================] - 2s 45us/step - loss: 0.0141 - val_loss: 0.0135\n",
      "Epoch 169/200\n",
      "35036/35036 [==============================] - 2s 48us/step - loss: 0.0131 - val_loss: 0.0179\n",
      "Epoch 170/200\n",
      "35036/35036 [==============================] - 2s 45us/step - loss: 0.0129 - val_loss: 0.0135\n",
      "Epoch 171/200\n",
      "35036/35036 [==============================] - 2s 45us/step - loss: 0.0142 - val_loss: 0.0150\n",
      "Epoch 172/200\n",
      "35036/35036 [==============================] - 2s 48us/step - loss: 0.0136 - val_loss: 0.0159\n",
      "Epoch 173/200\n",
      "35036/35036 [==============================] - 2s 47us/step - loss: 0.0130 - val_loss: 0.0188\n",
      "Epoch 174/200\n",
      "35036/35036 [==============================] - 2s 55us/step - loss: 0.0131 - val_loss: 0.0161\n",
      "Epoch 175/200\n",
      "35036/35036 [==============================] - 2s 47us/step - loss: 0.0138 - val_loss: 0.0150\n",
      "Epoch 176/200\n",
      "35036/35036 [==============================] - 2s 49us/step - loss: 0.0134 - val_loss: 0.0143\n",
      "Epoch 177/200\n",
      "35036/35036 [==============================] - 2s 62us/step - loss: 0.0127 - val_loss: 0.0153\n",
      "Epoch 178/200\n",
      "35036/35036 [==============================] - 2s 57us/step - loss: 0.0137 - val_loss: 0.0140\n",
      "Epoch 179/200\n",
      "35036/35036 [==============================] - 2s 50us/step - loss: 0.0129 - val_loss: 0.0136\n",
      "Epoch 180/200\n",
      "35036/35036 [==============================] - 2s 46us/step - loss: 0.0126 - val_loss: 0.0150\n",
      "Epoch 181/200\n",
      "35036/35036 [==============================] - 2s 44us/step - loss: 0.0130 - val_loss: 0.0142\n",
      "Epoch 182/200\n",
      "35036/35036 [==============================] - 2s 50us/step - loss: 0.0145 - val_loss: 0.0135\n",
      "Epoch 183/200\n",
      "35036/35036 [==============================] - 2s 47us/step - loss: 0.0131 - val_loss: 0.0141\n",
      "Epoch 184/200\n",
      "35036/35036 [==============================] - 2s 46us/step - loss: 0.0126 - val_loss: 0.0141\n",
      "Epoch 185/200\n",
      "35036/35036 [==============================] - 2s 44us/step - loss: 0.0129 - val_loss: 0.0140\n",
      "Epoch 186/200\n",
      "35036/35036 [==============================] - 1s 43us/step - loss: 0.0128 - val_loss: 0.0151\n",
      "Epoch 187/200\n",
      "35036/35036 [==============================] - 2s 43us/step - loss: 0.0135 - val_loss: 0.0147\n",
      "Epoch 188/200\n",
      "35036/35036 [==============================] - 2s 48us/step - loss: 0.0138 - val_loss: 0.0151\n",
      "Epoch 189/200\n",
      "35036/35036 [==============================] - 2s 45us/step - loss: 0.0133 - val_loss: 0.0143\n",
      "Epoch 190/200\n",
      "35036/35036 [==============================] - 2s 46us/step - loss: 0.0128 - val_loss: 0.0141\n",
      "Epoch 191/200\n",
      "35036/35036 [==============================] - 2s 44us/step - loss: 0.0133 - val_loss: 0.0147\n",
      "Epoch 192/200\n",
      "35036/35036 [==============================] - 2s 44us/step - loss: 0.0128 - val_loss: 0.0174\n",
      "Epoch 193/200\n",
      "35036/35036 [==============================] - 2s 46us/step - loss: 0.0132 - val_loss: 0.0135\n",
      "Epoch 194/200\n",
      "35036/35036 [==============================] - 2s 46us/step - loss: 0.0140 - val_loss: 0.0160\n",
      "Epoch 195/200\n",
      "35036/35036 [==============================] - 2s 47us/step - loss: 0.0133 - val_loss: 0.0153\n",
      "Epoch 196/200\n",
      "35036/35036 [==============================] - 2s 49us/step - loss: 0.0133 - val_loss: 0.0168\n",
      "Epoch 197/200\n",
      "35036/35036 [==============================] - 1s 43us/step - loss: 0.0129 - val_loss: 0.0172\n",
      "Epoch 198/200\n",
      "35036/35036 [==============================] - 2s 46us/step - loss: 0.0141 - val_loss: 0.0143\n",
      "Epoch 199/200\n",
      "35036/35036 [==============================] - 1s 41us/step - loss: 0.0130 - val_loss: 0.0134\n",
      "Epoch 200/200\n",
      "35036/35036 [==============================] - 1s 43us/step - loss: 0.0131 - val_loss: 0.0136\n"
     ]
    }
   ],
   "source": [
    "K.clear_session()\n",
    "model_lstm = Sequential()\n",
    "model_lstm.add(LSTM(68, input_shape=(1, dims),\n",
    "                    activation='relu',kernel_initializer='lecun_uniform',\n",
    "                    return_sequences=False,\n",
    "                    kernel_regularizer=regularizers.l2(0.01)))\n",
    "model_lstm.add(Dense(1))\n",
    "model_lstm.compile(loss='mean_squared_error', optimizer=Adam(lr=learning_rate))\n",
    "history_model_lstm = model_lstm.fit(X_tr_t,y_train_sc,epochs=training_epochs,\n",
    "                                    batch_size=batch,verbose=1,shuffle=True,\n",
    "                                    validation_data=(X_tst_t, y_test_sc))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x1a2ad36d30>"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(history.history['loss'], label='ANN_loss')\n",
    "plt.plot(history.history['val_loss'], label='ANN_Val_loss')\n",
    "plt.plot(history_model_lstm.history['loss'], label='LSTM_loss')\n",
    "plt.plot(history_model_lstm.history['val_loss'], label='LSTM_Val_loss')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x1189976d8>"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "y_pred_lstm = model_lstm.predict(X_tst_t)\n",
    "plt.plot(y_test_sc, label='True')\n",
    "plt.plot(y_pred_lstm, label='Predicted')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Predicting Multiple Values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "from sklearn.preprocessing import StandardScaler"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.read_csv('temp.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset = df[['year','month','day','hour','DEWP','TEMP','PRES','Iws','Is','Ir']]\n",
    "from collections import Counter\n",
    "cut = int(dataset.shape[0]*0.8)\n",
    "train = dataset.loc[:cut]\n",
    "test = dataset.loc[cut:]\n",
    "\n",
    "train = train.drop(['year'], axis=1)\n",
    "test = test.drop(['year'], axis=1)\n",
    "for s in range(1,25):\n",
    "    train['shift_{}'.format(s)] = train['TEMP'].shift(s)\n",
    "    test['shift_{}'.format(s)] = test['TEMP'].shift(s)\n",
    "\n",
    "for i in range(-1,-25,-1):\n",
    "    train['lag_{}'.format(i)] = train['TEMP'].shift(i)\n",
    "    test['lag_{}'.format(i)] = test['TEMP'].shift(i)\n",
    "\n",
    "train = train.drop('TEMP', axis=1)\n",
    "test = test.drop('TEMP', axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "target_vars = ['lag_-1','lag_-2','lag_-3','lag_-4',\n",
    "               'lag_-5','lag_-6','lag_-7','lag_-8',\n",
    "               'lag_-9','lag_-10','lag_-11','lag_-12',\n",
    "               'lag_-13','lag_-14','lag_-15','lag_-16',\n",
    "               'lag_-17','lag_-18','lag_-19','lag_-20',\n",
    "               'lag_-21','lag_-22','lag_-23','lag_-24']\n",
    "X_train = train.dropna().drop(target_vars, axis=1)\n",
    "y_train = train.dropna()[target_vars]\n",
    "\n",
    "X_test = test.dropna().drop(target_vars, axis=1)\n",
    "y_test = test.dropna()[target_vars]\n",
    "\n",
    "X_train_sc = StandardScaler().fit_transform(X_train)\n",
    "y_train_sc = StandardScaler().fit_transform(y_train)\n",
    "\n",
    "X_test_sc = StandardScaler().fit_transform(X_test)\n",
    "y_test_sc = StandardScaler().fit_transform(y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_tr_t = X_train_sc.reshape(X_train_sc.shape[0], 1, X_train_sc.shape[1])\n",
    "\n",
    "X_tst_t = X_test_sc.reshape(X_test_sc.shape[0], 1, X_test_sc.shape[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.models import Sequential\n",
    "from keras.layers import Dense\n",
    "from keras.layers import Dropout\n",
    "from keras.layers import LSTM\n",
    "from keras.optimizers import Adam\n",
    "from keras import regularizers\n",
    "import keras.backend as K"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Hyperparameter for Deep Learning\n",
    "learning_rate = 0.01\n",
    "training_epochs = 500\n",
    "display_step = 200\n",
    "factor = 0.10\n",
    "batch = 128\n",
    "seed = 7\n",
    "beta_1 = 0.9\n",
    "beta_2 = 0.999\n",
    "epsilon = None\n",
    "decay = 0.001\n",
    "np.random.seed(seed)\n",
    "dims = X_train_sc.shape[1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 35012 samples, validate on 8717 samples\n",
      "Epoch 1/500\n",
      "35012/35012 [==============================] - 2s 45us/step - loss: 0.2028 - val_loss: 0.0788\n",
      "Epoch 2/500\n",
      "35012/35012 [==============================] - 1s 25us/step - loss: 0.0702 - val_loss: 0.0821\n",
      "Epoch 3/500\n",
      "35012/35012 [==============================] - 1s 26us/step - loss: 0.0706 - val_loss: 0.0836\n",
      "Epoch 4/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0671 - val_loss: 0.0693\n",
      "Epoch 5/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0677 - val_loss: 0.0760\n",
      "Epoch 6/500\n",
      "35012/35012 [==============================] - 1s 26us/step - loss: 0.0662 - val_loss: 0.0743\n",
      "Epoch 7/500\n",
      "35012/35012 [==============================] - 1s 32us/step - loss: 0.0647 - val_loss: 0.0701\n",
      "Epoch 8/500\n",
      "35012/35012 [==============================] - 1s 26us/step - loss: 0.0658 - val_loss: 0.0749\n",
      "Epoch 9/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0645 - val_loss: 0.0744\n",
      "Epoch 10/500\n",
      "35012/35012 [==============================] - 1s 32us/step - loss: 0.0633 - val_loss: 0.0703\n",
      "Epoch 11/500\n",
      "35012/35012 [==============================] - 1s 25us/step - loss: 0.0639 - val_loss: 0.0740\n",
      "Epoch 12/500\n",
      "35012/35012 [==============================] - 1s 25us/step - loss: 0.0634 - val_loss: 0.0707\n",
      "Epoch 13/500\n",
      "35012/35012 [==============================] - 1s 25us/step - loss: 0.0632 - val_loss: 0.0679\n",
      "Epoch 14/500\n",
      "35012/35012 [==============================] - 1s 25us/step - loss: 0.0628 - val_loss: 0.0715\n",
      "Epoch 15/500\n",
      "35012/35012 [==============================] - 1s 26us/step - loss: 0.0628 - val_loss: 0.0776\n",
      "Epoch 16/500\n",
      "35012/35012 [==============================] - 1s 26us/step - loss: 0.0624 - val_loss: 0.0719\n",
      "Epoch 17/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0614 - val_loss: 0.0681\n",
      "Epoch 18/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0624 - val_loss: 0.0761\n",
      "Epoch 19/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0613 - val_loss: 0.0758\n",
      "Epoch 20/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0626 - val_loss: 0.0728\n",
      "Epoch 21/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0622 - val_loss: 0.0667\n",
      "Epoch 22/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0608 - val_loss: 0.0671\n",
      "Epoch 23/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0613 - val_loss: 0.0668\n",
      "Epoch 24/500\n",
      "35012/35012 [==============================] - 1s 26us/step - loss: 0.0604 - val_loss: 0.0678\n",
      "Epoch 25/500\n",
      "35012/35012 [==============================] - 1s 25us/step - loss: 0.0603 - val_loss: 0.0664\n",
      "Epoch 26/500\n",
      "35012/35012 [==============================] - 1s 26us/step - loss: 0.0607 - val_loss: 0.0695\n",
      "Epoch 27/500\n",
      "35012/35012 [==============================] - 1s 26us/step - loss: 0.0609 - val_loss: 0.0725\n",
      "Epoch 28/500\n",
      "35012/35012 [==============================] - 1s 26us/step - loss: 0.0610 - val_loss: 0.0677\n",
      "Epoch 29/500\n",
      "35012/35012 [==============================] - 1s 26us/step - loss: 0.0604 - val_loss: 0.0677\n",
      "Epoch 30/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0608 - val_loss: 0.0681\n",
      "Epoch 31/500\n",
      "35012/35012 [==============================] - 1s 26us/step - loss: 0.0600 - val_loss: 0.0701\n",
      "Epoch 32/500\n",
      "35012/35012 [==============================] - 1s 26us/step - loss: 0.0594 - val_loss: 0.0705\n",
      "Epoch 33/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0601 - val_loss: 0.0668\n",
      "Epoch 34/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0600 - val_loss: 0.0741\n",
      "Epoch 35/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0594 - val_loss: 0.0749\n",
      "Epoch 36/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0602 - val_loss: 0.0667\n",
      "Epoch 37/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0592 - val_loss: 0.0677\n",
      "Epoch 38/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0600 - val_loss: 0.0739\n",
      "Epoch 39/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0593 - val_loss: 0.0673\n",
      "Epoch 40/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0593 - val_loss: 0.0704\n",
      "Epoch 41/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0588 - val_loss: 0.0685\n",
      "Epoch 42/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0589 - val_loss: 0.0670\n",
      "Epoch 43/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0587 - val_loss: 0.0637\n",
      "Epoch 44/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0595 - val_loss: 0.0651\n",
      "Epoch 45/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0588 - val_loss: 0.0696\n",
      "Epoch 46/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0584 - val_loss: 0.0688\n",
      "Epoch 47/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0590 - val_loss: 0.0666\n",
      "Epoch 48/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0580 - val_loss: 0.0637\n",
      "Epoch 49/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0585 - val_loss: 0.0686\n",
      "Epoch 50/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0594 - val_loss: 0.0669\n",
      "Epoch 51/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0589 - val_loss: 0.0693\n",
      "Epoch 52/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0587 - val_loss: 0.0670\n",
      "Epoch 53/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0585 - val_loss: 0.0642\n",
      "Epoch 54/500\n",
      "35012/35012 [==============================] - 1s 32us/step - loss: 0.0583 - val_loss: 0.0667\n",
      "Epoch 55/500\n",
      "35012/35012 [==============================] - 1s 31us/step - loss: 0.0588 - val_loss: 0.0649\n",
      "Epoch 56/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0590 - val_loss: 0.0685\n",
      "Epoch 57/500\n",
      "35012/35012 [==============================] - 1s 31us/step - loss: 0.0585 - val_loss: 0.0719\n",
      "Epoch 58/500\n",
      "35012/35012 [==============================] - 1s 32us/step - loss: 0.0589 - val_loss: 0.0672\n",
      "Epoch 59/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0588 - val_loss: 0.0693\n",
      "Epoch 60/500\n",
      "35012/35012 [==============================] - 1s 31us/step - loss: 0.0585 - val_loss: 0.0651\n",
      "Epoch 61/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0589 - val_loss: 0.0651\n",
      "Epoch 62/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0584 - val_loss: 0.0705\n",
      "Epoch 63/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0586 - val_loss: 0.0685\n",
      "Epoch 64/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0581 - val_loss: 0.0672\n",
      "Epoch 65/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0593 - val_loss: 0.0654\n",
      "Epoch 66/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0584 - val_loss: 0.0685\n",
      "Epoch 67/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0586 - val_loss: 0.0687\n",
      "Epoch 68/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0587 - val_loss: 0.0664\n",
      "Epoch 69/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0586 - val_loss: 0.0660\n",
      "Epoch 70/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0586 - val_loss: 0.0650\n",
      "Epoch 71/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0588 - val_loss: 0.0655\n",
      "Epoch 72/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0586 - val_loss: 0.0651\n",
      "Epoch 73/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0587 - val_loss: 0.0639\n",
      "Epoch 74/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0585 - val_loss: 0.0662\n",
      "Epoch 75/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0585 - val_loss: 0.0672\n",
      "Epoch 76/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0585 - val_loss: 0.0703\n",
      "Epoch 77/500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0592 - val_loss: 0.0661\n",
      "Epoch 78/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0583 - val_loss: 0.0685\n",
      "Epoch 79/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0584 - val_loss: 0.0655\n",
      "Epoch 80/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0587 - val_loss: 0.0669\n",
      "Epoch 81/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0584 - val_loss: 0.0644\n",
      "Epoch 82/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0591 - val_loss: 0.0678\n",
      "Epoch 83/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0586 - val_loss: 0.0671\n",
      "Epoch 84/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0583 - val_loss: 0.0706\n",
      "Epoch 85/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0583 - val_loss: 0.0668\n",
      "Epoch 86/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0588 - val_loss: 0.0664\n",
      "Epoch 87/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0587 - val_loss: 0.0697\n",
      "Epoch 88/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0587 - val_loss: 0.0661\n",
      "Epoch 89/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0581 - val_loss: 0.0650\n",
      "Epoch 90/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0586 - val_loss: 0.0670\n",
      "Epoch 91/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0585 - val_loss: 0.0639\n",
      "Epoch 92/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0592 - val_loss: 0.0646\n",
      "Epoch 93/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0582 - val_loss: 0.0662\n",
      "Epoch 94/500\n",
      "35012/35012 [==============================] - 1s 26us/step - loss: 0.0582 - val_loss: 0.0625\n",
      "Epoch 95/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0589 - val_loss: 0.0663\n",
      "Epoch 96/500\n",
      "35012/35012 [==============================] - 1s 26us/step - loss: 0.0582 - val_loss: 0.0652\n",
      "Epoch 97/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0585 - val_loss: 0.0643\n",
      "Epoch 98/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0585 - val_loss: 0.0686\n",
      "Epoch 99/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0580 - val_loss: 0.0708\n",
      "Epoch 100/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0590 - val_loss: 0.0645\n",
      "Epoch 101/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0580 - val_loss: 0.0673\n",
      "Epoch 102/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0583 - val_loss: 0.0695\n",
      "Epoch 103/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0585 - val_loss: 0.0658\n",
      "Epoch 104/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0586 - val_loss: 0.0675\n",
      "Epoch 105/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0582 - val_loss: 0.0718\n",
      "Epoch 106/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0586 - val_loss: 0.0660\n",
      "Epoch 107/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0583 - val_loss: 0.0684\n",
      "Epoch 108/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0581 - val_loss: 0.0640\n",
      "Epoch 109/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0585 - val_loss: 0.0702\n",
      "Epoch 110/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0587 - val_loss: 0.0655\n",
      "Epoch 111/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0581 - val_loss: 0.0664\n",
      "Epoch 112/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0582 - val_loss: 0.0663\n",
      "Epoch 113/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0585 - val_loss: 0.0677\n",
      "Epoch 114/500\n",
      "35012/35012 [==============================] - 1s 31us/step - loss: 0.0579 - val_loss: 0.0675\n",
      "Epoch 115/500\n",
      "35012/35012 [==============================] - 1s 32us/step - loss: 0.0585 - val_loss: 0.0660\n",
      "Epoch 116/500\n",
      "35012/35012 [==============================] - 1s 32us/step - loss: 0.0584 - val_loss: 0.0645\n",
      "Epoch 117/500\n",
      "35012/35012 [==============================] - 1s 31us/step - loss: 0.0584 - val_loss: 0.0661\n",
      "Epoch 118/500\n",
      "35012/35012 [==============================] - 1s 31us/step - loss: 0.0585 - val_loss: 0.0644\n",
      "Epoch 119/500\n",
      "35012/35012 [==============================] - 1s 31us/step - loss: 0.0586 - val_loss: 0.0680\n",
      "Epoch 120/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0590 - val_loss: 0.0655\n",
      "Epoch 121/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0587 - val_loss: 0.0689\n",
      "Epoch 122/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0581 - val_loss: 0.0686\n",
      "Epoch 123/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0583 - val_loss: 0.0704\n",
      "Epoch 124/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0591 - val_loss: 0.0685\n",
      "Epoch 125/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0585 - val_loss: 0.0662\n",
      "Epoch 126/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0578 - val_loss: 0.0679\n",
      "Epoch 127/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0591 - val_loss: 0.0659\n",
      "Epoch 128/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0586 - val_loss: 0.0670\n",
      "Epoch 129/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0580 - val_loss: 0.0644\n",
      "Epoch 130/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0580 - val_loss: 0.0665\n",
      "Epoch 131/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0582 - val_loss: 0.0648\n",
      "Epoch 132/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0588 - val_loss: 0.0680\n",
      "Epoch 133/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0580 - val_loss: 0.0759\n",
      "Epoch 134/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0584 - val_loss: 0.0656\n",
      "Epoch 135/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0589 - val_loss: 0.0654\n",
      "Epoch 136/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0586 - val_loss: 0.0665\n",
      "Epoch 137/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0591 - val_loss: 0.0711\n",
      "Epoch 138/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0586 - val_loss: 0.0637\n",
      "Epoch 139/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0587 - val_loss: 0.0638\n",
      "Epoch 140/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0585 - val_loss: 0.0663\n",
      "Epoch 141/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0585 - val_loss: 0.0654\n",
      "Epoch 142/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0584 - val_loss: 0.0644\n",
      "Epoch 143/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0581 - val_loss: 0.0656\n",
      "Epoch 144/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0583 - val_loss: 0.0673\n",
      "Epoch 145/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0582 - val_loss: 0.0632\n",
      "Epoch 146/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0582 - val_loss: 0.0649\n",
      "Epoch 147/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0581 - val_loss: 0.0659\n",
      "Epoch 148/500\n",
      "35012/35012 [==============================] - 1s 32us/step - loss: 0.0585 - val_loss: 0.0652\n",
      "Epoch 149/500\n",
      "35012/35012 [==============================] - 1s 31us/step - loss: 0.0587 - val_loss: 0.0653\n",
      "Epoch 150/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0576 - val_loss: 0.0651\n",
      "Epoch 151/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0585 - val_loss: 0.0637\n",
      "Epoch 152/500\n",
      "35012/35012 [==============================] - 1s 31us/step - loss: 0.0585 - val_loss: 0.0673\n",
      "Epoch 153/500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0582 - val_loss: 0.0644\n",
      "Epoch 154/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0585 - val_loss: 0.0666\n",
      "Epoch 155/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0584 - val_loss: 0.0673\n",
      "Epoch 156/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0588 - val_loss: 0.0664\n",
      "Epoch 157/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0585 - val_loss: 0.0644\n",
      "Epoch 158/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0587 - val_loss: 0.0663\n",
      "Epoch 159/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0589 - val_loss: 0.0656\n",
      "Epoch 160/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0578 - val_loss: 0.0665\n",
      "Epoch 161/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0591 - val_loss: 0.0669\n",
      "Epoch 162/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0584 - val_loss: 0.0660\n",
      "Epoch 163/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0582 - val_loss: 0.0655\n",
      "Epoch 164/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0588 - val_loss: 0.0688\n",
      "Epoch 165/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0581 - val_loss: 0.0634\n",
      "Epoch 166/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0579 - val_loss: 0.0661\n",
      "Epoch 167/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0588 - val_loss: 0.0647\n",
      "Epoch 168/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0595 - val_loss: 0.0675\n",
      "Epoch 169/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0587 - val_loss: 0.0665\n",
      "Epoch 170/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0583 - val_loss: 0.0645\n",
      "Epoch 171/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0581 - val_loss: 0.0657\n",
      "Epoch 172/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0581 - val_loss: 0.0658\n",
      "Epoch 173/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0582 - val_loss: 0.0691\n",
      "Epoch 174/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0585 - val_loss: 0.0698\n",
      "Epoch 175/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0582 - val_loss: 0.0627\n",
      "Epoch 176/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0585 - val_loss: 0.0722\n",
      "Epoch 177/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0583 - val_loss: 0.0679\n",
      "Epoch 178/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0588 - val_loss: 0.0699\n",
      "Epoch 179/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0584 - val_loss: 0.0648\n",
      "Epoch 180/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0594 - val_loss: 0.0663\n",
      "Epoch 181/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0580 - val_loss: 0.0662\n",
      "Epoch 182/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0583 - val_loss: 0.0641\n",
      "Epoch 183/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0588 - val_loss: 0.0644\n",
      "Epoch 184/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0581 - val_loss: 0.0688\n",
      "Epoch 185/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0585 - val_loss: 0.0657\n",
      "Epoch 186/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0583 - val_loss: 0.0637\n",
      "Epoch 187/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0589 - val_loss: 0.0653\n",
      "Epoch 188/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0584 - val_loss: 0.0665\n",
      "Epoch 189/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0587 - val_loss: 0.0657\n",
      "Epoch 190/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0582 - val_loss: 0.0643\n",
      "Epoch 191/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0585 - val_loss: 0.0693\n",
      "Epoch 192/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0585 - val_loss: 0.0669\n",
      "Epoch 193/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0587 - val_loss: 0.0656\n",
      "Epoch 194/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0586 - val_loss: 0.0640\n",
      "Epoch 195/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0583 - val_loss: 0.0641\n",
      "Epoch 196/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0583 - val_loss: 0.0778\n",
      "Epoch 197/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0594 - val_loss: 0.0647\n",
      "Epoch 198/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0588 - val_loss: 0.0657\n",
      "Epoch 199/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0582 - val_loss: 0.0653\n",
      "Epoch 200/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0578 - val_loss: 0.0639\n",
      "Epoch 201/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0586 - val_loss: 0.0754\n",
      "Epoch 202/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0590 - val_loss: 0.0689\n",
      "Epoch 203/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0579 - val_loss: 0.0658\n",
      "Epoch 204/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0584 - val_loss: 0.0666\n",
      "Epoch 205/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0586 - val_loss: 0.0662\n",
      "Epoch 206/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0586 - val_loss: 0.0681\n",
      "Epoch 207/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0586 - val_loss: 0.0657\n",
      "Epoch 208/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0583 - val_loss: 0.0672\n",
      "Epoch 209/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0586 - val_loss: 0.0667\n",
      "Epoch 210/500\n",
      "35012/35012 [==============================] - 1s 31us/step - loss: 0.0585 - val_loss: 0.0644\n",
      "Epoch 211/500\n",
      "35012/35012 [==============================] - 1s 31us/step - loss: 0.0582 - val_loss: 0.0654\n",
      "Epoch 212/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0590 - val_loss: 0.0667\n",
      "Epoch 213/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0580 - val_loss: 0.0660\n",
      "Epoch 214/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0585 - val_loss: 0.0628\n",
      "Epoch 215/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0588 - val_loss: 0.0743\n",
      "Epoch 216/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0588 - val_loss: 0.0656\n",
      "Epoch 217/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0584 - val_loss: 0.0667\n",
      "Epoch 218/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0586 - val_loss: 0.0695\n",
      "Epoch 219/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0587 - val_loss: 0.0641\n",
      "Epoch 220/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0590 - val_loss: 0.0685\n",
      "Epoch 221/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0589 - val_loss: 0.0648\n",
      "Epoch 222/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0584 - val_loss: 0.0658\n",
      "Epoch 223/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0585 - val_loss: 0.0643\n",
      "Epoch 224/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0585 - val_loss: 0.0665\n",
      "Epoch 225/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0584 - val_loss: 0.0645\n",
      "Epoch 226/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0586 - val_loss: 0.0649\n",
      "Epoch 227/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0584 - val_loss: 0.0699\n",
      "Epoch 228/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0588 - val_loss: 0.0654\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 229/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0585 - val_loss: 0.0641\n",
      "Epoch 230/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0589 - val_loss: 0.0657\n",
      "Epoch 231/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0585 - val_loss: 0.0659\n",
      "Epoch 232/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0587 - val_loss: 0.0653\n",
      "Epoch 233/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0582 - val_loss: 0.0658\n",
      "Epoch 234/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0584 - val_loss: 0.0652\n",
      "Epoch 235/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0580 - val_loss: 0.0665\n",
      "Epoch 236/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0584 - val_loss: 0.0703\n",
      "Epoch 237/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0583 - val_loss: 0.0676\n",
      "Epoch 238/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0587 - val_loss: 0.0658\n",
      "Epoch 239/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0584 - val_loss: 0.0682\n",
      "Epoch 240/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0585 - val_loss: 0.0656\n",
      "Epoch 241/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0580 - val_loss: 0.0654\n",
      "Epoch 242/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0584 - val_loss: 0.0697\n",
      "Epoch 243/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0586 - val_loss: 0.0659\n",
      "Epoch 244/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0586 - val_loss: 0.0649\n",
      "Epoch 245/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0590 - val_loss: 0.0663\n",
      "Epoch 246/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0582 - val_loss: 0.0635\n",
      "Epoch 247/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0581 - val_loss: 0.0647\n",
      "Epoch 248/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0578 - val_loss: 0.0650\n",
      "Epoch 249/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0583 - val_loss: 0.0670\n",
      "Epoch 250/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0585 - val_loss: 0.0651\n",
      "Epoch 251/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0586 - val_loss: 0.0730\n",
      "Epoch 252/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0585 - val_loss: 0.0674\n",
      "Epoch 253/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0580 - val_loss: 0.0657\n",
      "Epoch 254/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0582 - val_loss: 0.0650\n",
      "Epoch 255/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0578 - val_loss: 0.0668\n",
      "Epoch 256/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0587 - val_loss: 0.0650\n",
      "Epoch 257/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0579 - val_loss: 0.0633\n",
      "Epoch 258/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0581 - val_loss: 0.0655\n",
      "Epoch 259/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0581 - val_loss: 0.0639\n",
      "Epoch 260/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0588 - val_loss: 0.0704\n",
      "Epoch 261/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0587 - val_loss: 0.0677\n",
      "Epoch 262/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0587 - val_loss: 0.0651\n",
      "Epoch 263/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0587 - val_loss: 0.0696\n",
      "Epoch 264/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0586 - val_loss: 0.0650\n",
      "Epoch 265/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0585 - val_loss: 0.0658\n",
      "Epoch 266/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0581 - val_loss: 0.0668\n",
      "Epoch 267/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0585 - val_loss: 0.0771\n",
      "Epoch 268/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0594 - val_loss: 0.0665\n",
      "Epoch 269/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0585 - val_loss: 0.0697\n",
      "Epoch 270/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0583 - val_loss: 0.0690\n",
      "Epoch 271/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0582 - val_loss: 0.0713\n",
      "Epoch 272/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0589 - val_loss: 0.0692\n",
      "Epoch 273/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0592 - val_loss: 0.0653\n",
      "Epoch 274/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0586 - val_loss: 0.0686\n",
      "Epoch 275/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0580 - val_loss: 0.0647\n",
      "Epoch 276/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0591 - val_loss: 0.0635\n",
      "Epoch 277/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0579 - val_loss: 0.0682\n",
      "Epoch 278/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0586 - val_loss: 0.0664\n",
      "Epoch 279/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0585 - val_loss: 0.0654\n",
      "Epoch 280/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0588 - val_loss: 0.0673\n",
      "Epoch 281/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0582 - val_loss: 0.0668\n",
      "Epoch 282/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0583 - val_loss: 0.0648\n",
      "Epoch 283/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0583 - val_loss: 0.0646\n",
      "Epoch 284/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0584 - val_loss: 0.0642\n",
      "Epoch 285/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0584 - val_loss: 0.0668\n",
      "Epoch 286/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0581 - val_loss: 0.0668\n",
      "Epoch 287/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0588 - val_loss: 0.0676\n",
      "Epoch 288/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0587 - val_loss: 0.0651\n",
      "Epoch 289/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0581 - val_loss: 0.0651\n",
      "Epoch 290/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0587 - val_loss: 0.0645\n",
      "Epoch 291/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0578 - val_loss: 0.0647\n",
      "Epoch 292/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0582 - val_loss: 0.0656\n",
      "Epoch 293/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0587 - val_loss: 0.0649\n",
      "Epoch 294/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0585 - val_loss: 0.0680\n",
      "Epoch 295/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0587 - val_loss: 0.0665\n",
      "Epoch 296/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0586 - val_loss: 0.0671\n",
      "Epoch 297/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0585 - val_loss: 0.0658\n",
      "Epoch 298/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0578 - val_loss: 0.0638\n",
      "Epoch 299/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0583 - val_loss: 0.0659\n",
      "Epoch 300/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0581 - val_loss: 0.0669\n",
      "Epoch 301/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0584 - val_loss: 0.0651\n",
      "Epoch 302/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0584 - val_loss: 0.0671\n",
      "Epoch 303/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0584 - val_loss: 0.0681\n",
      "Epoch 304/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0581 - val_loss: 0.0667\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 305/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0582 - val_loss: 0.0659\n",
      "Epoch 306/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0588 - val_loss: 0.0700\n",
      "Epoch 307/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0585 - val_loss: 0.0675\n",
      "Epoch 308/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0582 - val_loss: 0.0658\n",
      "Epoch 309/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0579 - val_loss: 0.0654\n",
      "Epoch 310/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0582 - val_loss: 0.0644\n",
      "Epoch 311/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0585 - val_loss: 0.0641\n",
      "Epoch 312/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0582 - val_loss: 0.0667\n",
      "Epoch 313/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0593 - val_loss: 0.0659\n",
      "Epoch 314/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0577 - val_loss: 0.0672\n",
      "Epoch 315/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0586 - val_loss: 0.0664\n",
      "Epoch 316/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0588 - val_loss: 0.0657\n",
      "Epoch 317/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0587 - val_loss: 0.0655\n",
      "Epoch 318/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0580 - val_loss: 0.0650\n",
      "Epoch 319/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0582 - val_loss: 0.0674\n",
      "Epoch 320/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0585 - val_loss: 0.0668\n",
      "Epoch 321/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0581 - val_loss: 0.0640\n",
      "Epoch 322/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0585 - val_loss: 0.0668\n",
      "Epoch 323/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0587 - val_loss: 0.0656\n",
      "Epoch 324/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0588 - val_loss: 0.0643\n",
      "Epoch 325/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0583 - val_loss: 0.0658\n",
      "Epoch 326/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0589 - val_loss: 0.0685\n",
      "Epoch 327/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0583 - val_loss: 0.0644\n",
      "Epoch 328/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0583 - val_loss: 0.0649\n",
      "Epoch 329/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0582 - val_loss: 0.0641\n",
      "Epoch 330/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0585 - val_loss: 0.0702\n",
      "Epoch 331/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0582 - val_loss: 0.0735\n",
      "Epoch 332/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0587 - val_loss: 0.0689\n",
      "Epoch 333/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0584 - val_loss: 0.0642\n",
      "Epoch 334/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0584 - val_loss: 0.0642\n",
      "Epoch 335/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0583 - val_loss: 0.0672\n",
      "Epoch 336/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0581 - val_loss: 0.0669\n",
      "Epoch 337/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0581 - val_loss: 0.0695\n",
      "Epoch 338/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0589 - val_loss: 0.0713\n",
      "Epoch 339/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0586 - val_loss: 0.0685\n",
      "Epoch 340/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0582 - val_loss: 0.0649\n",
      "Epoch 341/500\n",
      "35012/35012 [==============================] - 1s 31us/step - loss: 0.0586 - val_loss: 0.0647\n",
      "Epoch 342/500\n",
      "35012/35012 [==============================] - 1s 31us/step - loss: 0.0583 - val_loss: 0.0660\n",
      "Epoch 343/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0581 - val_loss: 0.0638\n",
      "Epoch 344/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0587 - val_loss: 0.0659\n",
      "Epoch 345/500\n",
      "35012/35012 [==============================] - 1s 31us/step - loss: 0.0582 - val_loss: 0.0654\n",
      "Epoch 346/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0586 - val_loss: 0.0686\n",
      "Epoch 347/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0583 - val_loss: 0.0647\n",
      "Epoch 348/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0585 - val_loss: 0.0715\n",
      "Epoch 349/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0588 - val_loss: 0.0650\n",
      "Epoch 350/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0585 - val_loss: 0.0653\n",
      "Epoch 351/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0588 - val_loss: 0.0680\n",
      "Epoch 352/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0583 - val_loss: 0.0746\n",
      "Epoch 353/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0583 - val_loss: 0.0669\n",
      "Epoch 354/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0587 - val_loss: 0.0694\n",
      "Epoch 355/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0591 - val_loss: 0.0666\n",
      "Epoch 356/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0579 - val_loss: 0.0684\n",
      "Epoch 357/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0590 - val_loss: 0.0666\n",
      "Epoch 358/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0583 - val_loss: 0.0686\n",
      "Epoch 359/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0586 - val_loss: 0.0658\n",
      "Epoch 360/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0585 - val_loss: 0.0659\n",
      "Epoch 361/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0581 - val_loss: 0.0646\n",
      "Epoch 362/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0581 - val_loss: 0.0649\n",
      "Epoch 363/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0588 - val_loss: 0.0703\n",
      "Epoch 364/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0582 - val_loss: 0.0641\n",
      "Epoch 365/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0578 - val_loss: 0.0640\n",
      "Epoch 366/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0586 - val_loss: 0.0644\n",
      "Epoch 367/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0583 - val_loss: 0.0664\n",
      "Epoch 368/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0579 - val_loss: 0.0661\n",
      "Epoch 369/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0585 - val_loss: 0.0661\n",
      "Epoch 370/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0577 - val_loss: 0.0676\n",
      "Epoch 371/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0585 - val_loss: 0.0638\n",
      "Epoch 372/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0588 - val_loss: 0.0676\n",
      "Epoch 373/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0590 - val_loss: 0.0644\n",
      "Epoch 374/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0581 - val_loss: 0.0654\n",
      "Epoch 375/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0577 - val_loss: 0.0651\n",
      "Epoch 376/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0586 - val_loss: 0.0667\n",
      "Epoch 377/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0584 - val_loss: 0.0695\n",
      "Epoch 378/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0580 - val_loss: 0.0674\n",
      "Epoch 379/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0580 - val_loss: 0.0658\n",
      "Epoch 380/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0588 - val_loss: 0.0649\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 381/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0595 - val_loss: 0.0655\n",
      "Epoch 382/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0579 - val_loss: 0.0652\n",
      "Epoch 383/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0585 - val_loss: 0.0649\n",
      "Epoch 384/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0583 - val_loss: 0.0655\n",
      "Epoch 385/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0580 - val_loss: 0.0705\n",
      "Epoch 386/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0585 - val_loss: 0.0660\n",
      "Epoch 387/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0586 - val_loss: 0.0660\n",
      "Epoch 388/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0588 - val_loss: 0.0646\n",
      "Epoch 389/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0580 - val_loss: 0.0643\n",
      "Epoch 390/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0587 - val_loss: 0.0640\n",
      "Epoch 391/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0581 - val_loss: 0.0665\n",
      "Epoch 392/500\n",
      "35012/35012 [==============================] - 1s 31us/step - loss: 0.0584 - val_loss: 0.0639\n",
      "Epoch 393/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0585 - val_loss: 0.0638\n",
      "Epoch 394/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0580 - val_loss: 0.0668\n",
      "Epoch 395/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0588 - val_loss: 0.0657\n",
      "Epoch 396/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0582 - val_loss: 0.0688\n",
      "Epoch 397/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0578 - val_loss: 0.0679\n",
      "Epoch 398/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0582 - val_loss: 0.0642\n",
      "Epoch 399/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0587 - val_loss: 0.0638\n",
      "Epoch 400/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0591 - val_loss: 0.0700\n",
      "Epoch 401/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0579 - val_loss: 0.0667\n",
      "Epoch 402/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0585 - val_loss: 0.0654\n",
      "Epoch 403/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0586 - val_loss: 0.0655\n",
      "Epoch 404/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0588 - val_loss: 0.0660\n",
      "Epoch 405/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0599 - val_loss: 0.0655\n",
      "Epoch 406/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0581 - val_loss: 0.0637\n",
      "Epoch 407/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0590 - val_loss: 0.0676\n",
      "Epoch 408/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0581 - val_loss: 0.0695\n",
      "Epoch 409/500\n",
      "35012/35012 [==============================] - 1s 27us/step - loss: 0.0586 - val_loss: 0.0655\n",
      "Epoch 410/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0581 - val_loss: 0.0648\n",
      "Epoch 411/500\n",
      "35012/35012 [==============================] - 1s 28us/step - loss: 0.0586 - val_loss: 0.0676\n",
      "Epoch 412/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0586 - val_loss: 0.0734\n",
      "Epoch 413/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0579 - val_loss: 0.0667\n",
      "Epoch 414/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0583 - val_loss: 0.0677\n",
      "Epoch 415/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0585 - val_loss: 0.0708\n",
      "Epoch 416/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0584 - val_loss: 0.0665\n",
      "Epoch 417/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0580 - val_loss: 0.0664\n",
      "Epoch 418/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0585 - val_loss: 0.0686\n",
      "Epoch 419/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0587 - val_loss: 0.0637\n",
      "Epoch 420/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0582 - val_loss: 0.0686\n",
      "Epoch 421/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0582 - val_loss: 0.0657\n",
      "Epoch 422/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0581 - val_loss: 0.0643\n",
      "Epoch 423/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0580 - val_loss: 0.0657\n",
      "Epoch 424/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0584 - val_loss: 0.0727\n",
      "Epoch 425/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0582 - val_loss: 0.0656\n",
      "Epoch 426/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0580 - val_loss: 0.0689\n",
      "Epoch 427/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0590 - val_loss: 0.0681\n",
      "Epoch 428/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0586 - val_loss: 0.0658\n",
      "Epoch 429/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0586 - val_loss: 0.0649\n",
      "Epoch 430/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0588 - val_loss: 0.0657\n",
      "Epoch 431/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0580 - val_loss: 0.0638\n",
      "Epoch 432/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0589 - val_loss: 0.0683\n",
      "Epoch 433/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0587 - val_loss: 0.0695\n",
      "Epoch 434/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0589 - val_loss: 0.0659\n",
      "Epoch 435/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0584 - val_loss: 0.0644\n",
      "Epoch 436/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0586 - val_loss: 0.0669\n",
      "Epoch 437/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0593 - val_loss: 0.0668\n",
      "Epoch 438/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0581 - val_loss: 0.0660\n",
      "Epoch 439/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0587 - val_loss: 0.0682\n",
      "Epoch 440/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0587 - val_loss: 0.0662\n",
      "Epoch 441/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0581 - val_loss: 0.0647\n",
      "Epoch 442/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0591 - val_loss: 0.0647\n",
      "Epoch 443/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0584 - val_loss: 0.0650\n",
      "Epoch 444/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0578 - val_loss: 0.0652\n",
      "Epoch 445/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0582 - val_loss: 0.0663\n",
      "Epoch 446/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0584 - val_loss: 0.0715\n",
      "Epoch 447/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0594 - val_loss: 0.0646\n",
      "Epoch 448/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0586 - val_loss: 0.0659\n",
      "Epoch 449/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0588 - val_loss: 0.0660\n",
      "Epoch 450/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0588 - val_loss: 0.0645\n",
      "Epoch 451/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0580 - val_loss: 0.0648\n",
      "Epoch 452/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0582 - val_loss: 0.0655\n",
      "Epoch 453/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0583 - val_loss: 0.0668\n",
      "Epoch 454/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0587 - val_loss: 0.0645\n",
      "Epoch 455/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0586 - val_loss: 0.0657\n",
      "Epoch 456/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0582 - val_loss: 0.0685\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 457/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0589 - val_loss: 0.0633\n",
      "Epoch 458/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0587 - val_loss: 0.0730\n",
      "Epoch 459/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0586 - val_loss: 0.0656\n",
      "Epoch 460/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0581 - val_loss: 0.0657\n",
      "Epoch 461/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0581 - val_loss: 0.0666\n",
      "Epoch 462/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0591 - val_loss: 0.0658\n",
      "Epoch 463/500\n",
      "35012/35012 [==============================] - 1s 32us/step - loss: 0.0582 - val_loss: 0.0671\n",
      "Epoch 464/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0586 - val_loss: 0.0645\n",
      "Epoch 465/500\n",
      "35012/35012 [==============================] - 1s 31us/step - loss: 0.0585 - val_loss: 0.0681\n",
      "Epoch 466/500\n",
      "35012/35012 [==============================] - 1s 35us/step - loss: 0.0580 - val_loss: 0.0717\n",
      "Epoch 467/500\n",
      "35012/35012 [==============================] - 1s 34us/step - loss: 0.0586 - val_loss: 0.0700\n",
      "Epoch 468/500\n",
      "35012/35012 [==============================] - 1s 34us/step - loss: 0.0586 - val_loss: 0.0658\n",
      "Epoch 469/500\n",
      "35012/35012 [==============================] - 1s 34us/step - loss: 0.0583 - val_loss: 0.0651\n",
      "Epoch 470/500\n",
      "35012/35012 [==============================] - 1s 35us/step - loss: 0.0584 - val_loss: 0.0645\n",
      "Epoch 471/500\n",
      "35012/35012 [==============================] - 1s 33us/step - loss: 0.0583 - val_loss: 0.0653\n",
      "Epoch 472/500\n",
      "35012/35012 [==============================] - 1s 32us/step - loss: 0.0590 - val_loss: 0.0687\n",
      "Epoch 473/500\n",
      "35012/35012 [==============================] - 1s 32us/step - loss: 0.0587 - val_loss: 0.0693\n",
      "Epoch 474/500\n",
      "35012/35012 [==============================] - 1s 31us/step - loss: 0.0585 - val_loss: 0.0644\n",
      "Epoch 475/500\n",
      "35012/35012 [==============================] - 1s 31us/step - loss: 0.0587 - val_loss: 0.0653\n",
      "Epoch 476/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0582 - val_loss: 0.0674\n",
      "Epoch 477/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0581 - val_loss: 0.0653\n",
      "Epoch 478/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0585 - val_loss: 0.0737\n",
      "Epoch 479/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0586 - val_loss: 0.0746\n",
      "Epoch 480/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0588 - val_loss: 0.0639\n",
      "Epoch 481/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0582 - val_loss: 0.0647\n",
      "Epoch 482/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0587 - val_loss: 0.0686\n",
      "Epoch 483/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0580 - val_loss: 0.0680\n",
      "Epoch 484/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0583 - val_loss: 0.0646\n",
      "Epoch 485/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0582 - val_loss: 0.0658\n",
      "Epoch 486/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0582 - val_loss: 0.0641\n",
      "Epoch 487/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0584 - val_loss: 0.0664\n",
      "Epoch 488/500\n",
      "35012/35012 [==============================] - 1s 32us/step - loss: 0.0585 - val_loss: 0.0685\n",
      "Epoch 489/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0585 - val_loss: 0.0648\n",
      "Epoch 490/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0586 - val_loss: 0.0655\n",
      "Epoch 491/500\n",
      "35012/35012 [==============================] - 1s 31us/step - loss: 0.0585 - val_loss: 0.0687\n",
      "Epoch 492/500\n",
      "35012/35012 [==============================] - 1s 31us/step - loss: 0.0586 - val_loss: 0.0670\n",
      "Epoch 493/500\n",
      "35012/35012 [==============================] - 1s 31us/step - loss: 0.0581 - val_loss: 0.0686\n",
      "Epoch 494/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0590 - val_loss: 0.0642\n",
      "Epoch 495/500\n",
      "35012/35012 [==============================] - 1s 31us/step - loss: 0.0581 - val_loss: 0.0653\n",
      "Epoch 496/500\n",
      "35012/35012 [==============================] - 1s 31us/step - loss: 0.0593 - val_loss: 0.0662\n",
      "Epoch 497/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0576 - val_loss: 0.0684\n",
      "Epoch 498/500\n",
      "35012/35012 [==============================] - 1s 30us/step - loss: 0.0587 - val_loss: 0.0659\n",
      "Epoch 499/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0590 - val_loss: 0.0636\n",
      "Epoch 500/500\n",
      "35012/35012 [==============================] - 1s 29us/step - loss: 0.0587 - val_loss: 0.0705\n"
     ]
    }
   ],
   "source": [
    "K.clear_session()\n",
    "model_lag = Sequential()\n",
    "model_lag.add(LSTM(68, input_shape=(1, dims),\n",
    "                    activation='relu',kernel_initializer='lecun_uniform',\n",
    "                    return_sequences=False,\n",
    "                    kernel_regularizer=regularizers.l2(0.01)))\n",
    "model_lag.add(Dense(y_train_sc.shape[1]))\n",
    "model_lag.compile(loss='mean_squared_error', optimizer=Adam(lr=learning_rate))\n",
    "history_model_lag = model_lag.fit(X_tr_t,y_train_sc,epochs=training_epochs,\n",
    "                                    batch_size=batch,verbose=1,shuffle=True,\n",
    "                                    validation_data=(X_tst_t, y_test_sc))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x1189c4d30>]"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(history_model_lag.history['loss'])\n",
    "plt.plot(history_model_lag.history['val_loss'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "pred= model_lag.predict(X_tr_t)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x504 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(15,7))\n",
    "plt.plot(pred, 'r--')\n",
    "plt.plot(y_train_sc)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "beg=550\n",
    "\n",
    "\n",
    "selected_input = X_tst_t[beg:beg+1,:,:]\n",
    "real_output = y_test_sc[beg:beg+1]\n",
    "predicted_output = model_lag.predict(selected_input)\n",
    "\n",
    "\n",
    "\n",
    "predicted_output=np.transpose(predicted_output)\n",
    "real_output=np.transpose(real_output)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x504 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(15,7))\n",
    "plt.plot(predicted_output,'r')\n",
    "plt.plot(real_output, 'b')\n",
    "plt.show()"
   ]
  },
  {
   "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": "3.6.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}