notebook.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2.1.0\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import tensorflow as tf\n",
    "\n",
    "print(tf.__version__)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "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>Jour</th>\n",
       "      <th>Visites</th>\n",
       "      <th>Visiteurs uniques</th>\n",
       "      <th>Pages vues</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>01/01/2012</td>\n",
       "      <td>906</td>\n",
       "      <td>807</td>\n",
       "      <td>2184</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>02/01/2012</td>\n",
       "      <td>1655</td>\n",
       "      <td>1404</td>\n",
       "      <td>4234</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>03/01/2012</td>\n",
       "      <td>1728</td>\n",
       "      <td>1484</td>\n",
       "      <td>4574</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>04/01/2012</td>\n",
       "      <td>1604</td>\n",
       "      <td>1370</td>\n",
       "      <td>4123</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>05/01/2012</td>\n",
       "      <td>1885</td>\n",
       "      <td>1578</td>\n",
       "      <td>4663</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         Jour  Visites  Visiteurs uniques  Pages vues\n",
       "0  01/01/2012      906                807        2184\n",
       "1  02/01/2012     1655               1404        4234\n",
       "2  03/01/2012     1728               1484        4574\n",
       "3  04/01/2012     1604               1370        4123\n",
       "4  05/01/2012     1885               1578        4663"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dataset_path = os.path.join(\"dataset\", \"35288-STMALO-Statistiques-Web-Quotidien.csv\")\n",
    "df = pd.read_csv(dataset_path, sep=\";\")\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7f14a4e332e8>"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "df.plot(\"Jour\", \"Visiteurs uniques\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def normalize_series(series):\n",
    "    mean_series = np.mean(series)\n",
    "    max_series = np.max(series)\n",
    "    normalized_series = (series - mean_series)/max_series\n",
    "    return normalized_series, mean_series, max_series"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def unnormalize_series(series, mean_series, max_series):\n",
    "    return series*max_series + mean_series"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_series(time, series, format=\"-\", start=0, end=None):\n",
    "    plt.plot(time[start:end], series[start:end], format)\n",
    "    plt.xlabel(\"Time\")\n",
    "    plt.ylabel(\"Value\")\n",
    "    plt.grid(True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "def windowed_dataset(series, window_size, batch_size, shuffle_buffer):\n",
    "    series = tf.expand_dims(series, axis=-1)\n",
    "    ds = tf.data.Dataset.from_tensor_slices(series)\n",
    "    ds = ds.window(window_size + 1, shift=1, drop_remainder=True)\n",
    "    ds = ds.flat_map(lambda w: w.batch(window_size + 1))\n",
    "    ds = ds.shuffle(shuffle_buffer)\n",
    "    ds = ds.map(lambda w: (w[:-1], w[-1]))\n",
    "    ds = ds.batch(batch_size).prefetch(1)\n",
    "    return ds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "def model_forecast(model, series, window_size):\n",
    "    ds = tf.data.Dataset.from_tensor_slices(series)\n",
    "    ds = ds.window(window_size, shift=1, drop_remainder=True)\n",
    "    ds = ds.flat_map(lambda w: w.batch(window_size))\n",
    "    ds = ds.batch(32).prefetch(1)\n",
    "    forecast = model.predict(ds)\n",
    "    return forecast"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "There are 1096 days in the dataset.\n",
      "Split time : 767\n"
     ]
    }
   ],
   "source": [
    "series = np.array(df[\"Visites\"])\n",
    "dates = df[\"Jour\"]\n",
    "len_time = len(dates)\n",
    "time = np.arange(len_time)\n",
    "split_time = int(len_time*0.7)\n",
    "\n",
    "print(\"There are {} days in the dataset.\".format(len_time))\n",
    "print(\"Split time : {}\".format(split_time))\n",
    "\n",
    "# Normalize series\n",
    "series, mean_series, max_series  = normalize_series(series)\n",
    "\n",
    "# Train/Validation split\n",
    "time_train = time[:split_time]\n",
    "x_train = series[:split_time]\n",
    "time_valid = time[split_time:]\n",
    "x_valid = series[split_time:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/400\n",
      "6/6 [==============================] - 3s 511ms/step - loss: 0.0038 - mae: 0.0621\n",
      "Epoch 2/400\n",
      "6/6 [==============================] - 0s 26ms/step - loss: 0.0034 - mae: 0.0567\n",
      "Epoch 3/400\n",
      "6/6 [==============================] - 0s 26ms/step - loss: 0.0030 - mae: 0.0528\n",
      "Epoch 4/400\n",
      "6/6 [==============================] - 0s 27ms/step - loss: 0.0026 - mae: 0.0491\n",
      "Epoch 5/400\n",
      "6/6 [==============================] - 0s 31ms/step - loss: 0.0023 - mae: 0.0442\n",
      "Epoch 6/400\n",
      "6/6 [==============================] - ETA: 0s - loss: 0.0022 - mae: 0.041 - 0s 34ms/step - loss: 0.0021 - mae: 0.0414\n",
      "Epoch 7/400\n",
      "6/6 [==============================] - 0s 29ms/step - loss: 0.0020 - mae: 0.0394\n",
      "Epoch 8/400\n",
      "6/6 [==============================] - 0s 26ms/step - loss: 0.0019 - mae: 0.0384\n",
      "Epoch 9/400\n",
      "6/6 [==============================] - 0s 31ms/step - loss: 0.0018 - mae: 0.0378\n",
      "Epoch 10/400\n",
      "6/6 [==============================] - 0s 39ms/step - loss: 0.0018 - mae: 0.0375\n",
      "Epoch 11/400\n",
      "6/6 [==============================] - 0s 27ms/step - loss: 0.0018 - mae: 0.0375\n",
      "Epoch 12/400\n",
      "6/6 [==============================] - 0s 30ms/step - loss: 0.0018 - mae: 0.0375\n",
      "Epoch 13/400\n",
      "6/6 [==============================] - 0s 31ms/step - loss: 0.0017 - mae: 0.0376\n",
      "Epoch 14/400\n",
      "6/6 [==============================] - 0s 25ms/step - loss: 0.0018 - mae: 0.0371\n",
      "Epoch 15/400\n",
      "6/6 [==============================] - 0s 25ms/step - loss: 0.0017 - mae: 0.0372\n",
      "Epoch 16/400\n",
      "6/6 [==============================] - 0s 24ms/step - loss: 0.0017 - mae: 0.0369\n",
      "Epoch 17/400\n",
      "6/6 [==============================] - 0s 33ms/step - loss: 0.0017 - mae: 0.0365\n",
      "Epoch 18/400\n",
      "6/6 [==============================] - 0s 27ms/step - loss: 0.0017 - mae: 0.0366\n",
      "Epoch 19/400\n",
      "6/6 [==============================] - 0s 29ms/step - loss: 0.0017 - mae: 0.0365\n",
      "Epoch 20/400\n",
      "6/6 [==============================] - 0s 31ms/step - loss: 0.0017 - mae: 0.0362\n",
      "Epoch 21/400\n",
      "6/6 [==============================] - 0s 38ms/step - loss: 0.0017 - mae: 0.0363\n",
      "Epoch 22/400\n",
      "6/6 [==============================] - 0s 34ms/step - loss: 0.0017 - mae: 0.0361\n",
      "Epoch 23/400\n",
      "6/6 [==============================] - 0s 39ms/step - loss: 0.0017 - mae: 0.0359\n",
      "Epoch 24/400\n",
      "6/6 [==============================] - 0s 46ms/step - loss: 0.0017 - mae: 0.0358\n",
      "Epoch 25/400\n",
      "6/6 [==============================] - 0s 34ms/step - loss: 0.0017 - mae: 0.0357\n",
      "Epoch 26/400\n",
      "6/6 [==============================] - 0s 36ms/step - loss: 0.0016 - mae: 0.0357\n",
      "Epoch 27/400\n",
      "6/6 [==============================] - 0s 32ms/step - loss: 0.0016 - mae: 0.0356\n",
      "Epoch 28/400\n",
      "6/6 [==============================] - 0s 31ms/step - loss: 0.0016 - mae: 0.0352\n",
      "Epoch 29/400\n",
      "6/6 [==============================] - 0s 26ms/step - loss: 0.0016 - mae: 0.0354\n",
      "Epoch 30/400\n",
      "6/6 [==============================] - 0s 30ms/step - loss: 0.0016 - mae: 0.0352\n",
      "Epoch 31/400\n",
      "6/6 [==============================] - 0s 38ms/step - loss: 0.0016 - mae: 0.0350\n",
      "Epoch 32/400\n",
      "6/6 [==============================] - 0s 34ms/step - loss: 0.0016 - mae: 0.0350\n",
      "Epoch 33/400\n",
      "6/6 [==============================] - 0s 33ms/step - loss: 0.0016 - mae: 0.0346\n",
      "Epoch 34/400\n",
      "6/6 [==============================] - 0s 30ms/step - loss: 0.0016 - mae: 0.0346\n",
      "Epoch 35/400\n",
      "6/6 [==============================] - 0s 33ms/step - loss: 0.0016 - mae: 0.0345\n",
      "Epoch 36/400\n",
      "6/6 [==============================] - 0s 33ms/step - loss: 0.0016 - mae: 0.0343\n",
      "Epoch 37/400\n",
      "6/6 [==============================] - 0s 33ms/step - loss: 0.0016 - mae: 0.0342\n",
      "Epoch 38/400\n",
      "6/6 [==============================] - 0s 40ms/step - loss: 0.0015 - mae: 0.0343\n",
      "Epoch 39/400\n",
      "6/6 [==============================] - 0s 33ms/step - loss: 0.0016 - mae: 0.0340\n",
      "Epoch 40/400\n",
      "6/6 [==============================] - 0s 34ms/step - loss: 0.0015 - mae: 0.0337\n",
      "Epoch 41/400\n",
      "6/6 [==============================] - 0s 41ms/step - loss: 0.0015 - mae: 0.0339\n",
      "Epoch 42/400\n",
      "6/6 [==============================] - 0s 37ms/step - loss: 0.0015 - mae: 0.0336\n",
      "Epoch 43/400\n",
      "6/6 [==============================] - 0s 39ms/step - loss: 0.0015 - mae: 0.0334\n",
      "Epoch 44/400\n",
      "6/6 [==============================] - 0s 44ms/step - loss: 0.0015 - mae: 0.0333\n",
      "Epoch 45/400\n",
      "6/6 [==============================] - 0s 36ms/step - loss: 0.0015 - mae: 0.0333\n",
      "Epoch 46/400\n",
      "6/6 [==============================] - 0s 32ms/step - loss: 0.0015 - mae: 0.0329\n",
      "Epoch 47/400\n",
      "6/6 [==============================] - 0s 35ms/step - loss: 0.0015 - mae: 0.0333\n",
      "Epoch 48/400\n",
      "6/6 [==============================] - 0s 34ms/step - loss: 0.0015 - mae: 0.0327\n",
      "Epoch 49/400\n",
      "6/6 [==============================] - 0s 34ms/step - loss: 0.0015 - mae: 0.0331\n",
      "Epoch 50/400\n",
      "6/6 [==============================] - 0s 36ms/step - loss: 0.0015 - mae: 0.0325\n",
      "Epoch 51/400\n",
      "6/6 [==============================] - 0s 32ms/step - loss: 0.0015 - mae: 0.0327\n",
      "Epoch 52/400\n",
      "6/6 [==============================] - 0s 33ms/step - loss: 0.0014 - mae: 0.0323\n",
      "Epoch 53/400\n",
      "6/6 [==============================] - 0s 35ms/step - loss: 0.0014 - mae: 0.0324\n",
      "Epoch 54/400\n",
      "6/6 [==============================] - 0s 32ms/step - loss: 0.0014 - mae: 0.0321\n",
      "Epoch 55/400\n",
      "6/6 [==============================] - 0s 31ms/step - loss: 0.0014 - mae: 0.0325\n",
      "Epoch 56/400\n",
      "6/6 [==============================] - 0s 33ms/step - loss: 0.0014 - mae: 0.0320\n",
      "Epoch 57/400\n",
      "6/6 [==============================] - 0s 37ms/step - loss: 0.0014 - mae: 0.0326\n",
      "Epoch 58/400\n",
      "6/6 [==============================] - 0s 34ms/step - loss: 0.0014 - mae: 0.0321\n",
      "Epoch 59/400\n",
      "6/6 [==============================] - 0s 34ms/step - loss: 0.0014 - mae: 0.0319\n",
      "Epoch 60/400\n",
      "6/6 [==============================] - 0s 21ms/step - loss: 0.0014 - mae: 0.0321\n",
      "Epoch 61/400\n",
      "6/6 [==============================] - 0s 21ms/step - loss: 0.0014 - mae: 0.0312\n",
      "Epoch 62/400\n",
      "6/6 [==============================] - 0s 31ms/step - loss: 0.0014 - mae: 0.0316\n",
      "Epoch 63/400\n",
      "6/6 [==============================] - 0s 28ms/step - loss: 0.0013 - mae: 0.0313\n",
      "Epoch 64/400\n",
      "6/6 [==============================] - 0s 27ms/step - loss: 0.0013 - mae: 0.0311\n",
      "Epoch 65/400\n",
      "6/6 [==============================] - 0s 41ms/step - loss: 0.0013 - mae: 0.0316\n",
      "Epoch 66/400\n",
      "6/6 [==============================] - 0s 34ms/step - loss: 0.0013 - mae: 0.0312\n",
      "Epoch 67/400\n",
      "6/6 [==============================] - 0s 37ms/step - loss: 0.0013 - mae: 0.0312\n",
      "Epoch 68/400\n",
      "6/6 [==============================] - 0s 42ms/step - loss: 0.0013 - mae: 0.0310\n",
      "Epoch 69/400\n",
      "6/6 [==============================] - 0s 31ms/step - loss: 0.0013 - mae: 0.0308\n",
      "Epoch 70/400\n",
      "6/6 [==============================] - 0s 35ms/step - loss: 0.0013 - mae: 0.0309\n",
      "Epoch 71/400\n",
      "6/6 [==============================] - 0s 28ms/step - loss: 0.0013 - mae: 0.0307\n",
      "Epoch 72/400\n",
      "6/6 [==============================] - 0s 26ms/step - loss: 0.0013 - mae: 0.0308\n",
      "Epoch 73/400\n",
      "6/6 [==============================] - 0s 28ms/step - loss: 0.0013 - mae: 0.0304\n",
      "Epoch 74/400\n",
      "6/6 [==============================] - 0s 37ms/step - loss: 0.0013 - mae: 0.0310\n",
      "Epoch 75/400\n",
      "6/6 [==============================] - 0s 36ms/step - loss: 0.0013 - mae: 0.0303\n",
      "Epoch 76/400\n",
      "6/6 [==============================] - 0s 32ms/step - loss: 0.0013 - mae: 0.0305\n",
      "Epoch 77/400\n",
      "6/6 [==============================] - 0s 37ms/step - loss: 0.0013 - mae: 0.0304\n",
      "Epoch 78/400\n",
      "6/6 [==============================] - 0s 30ms/step - loss: 0.0013 - mae: 0.0303\n",
      "Epoch 79/400\n",
      "6/6 [==============================] - 0s 32ms/step - loss: 0.0013 - mae: 0.0315\n",
      "Epoch 80/400\n",
      "6/6 [==============================] - 0s 35ms/step - loss: 0.0013 - mae: 0.0302\n",
      "Epoch 81/400\n",
      "6/6 [==============================] - 0s 23ms/step - loss: 0.0012 - mae: 0.0304\n",
      "Epoch 82/400\n",
      "6/6 [==============================] - 0s 31ms/step - loss: 0.0012 - mae: 0.0300\n",
      "Epoch 83/400\n",
      "6/6 [==============================] - 0s 27ms/step - loss: 0.0012 - mae: 0.0303\n",
      "Epoch 84/400\n",
      "6/6 [==============================] - 0s 25ms/step - loss: 0.0012 - mae: 0.0300\n",
      "Epoch 85/400\n",
      "6/6 [==============================] - 0s 27ms/step - loss: 0.0012 - mae: 0.0304\n",
      "Epoch 86/400\n",
      "6/6 [==============================] - 0s 22ms/step - loss: 0.0012 - mae: 0.0300\n",
      "Epoch 87/400\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "6/6 [==============================] - 0s 20ms/step - loss: 0.0012 - mae: 0.0302\n",
      "Epoch 88/400\n",
      "6/6 [==============================] - 0s 21ms/step - loss: 0.0012 - mae: 0.0300\n",
      "Epoch 89/400\n",
      "6/6 [==============================] - 0s 21ms/step - loss: 0.0012 - mae: 0.0299\n",
      "Epoch 90/400\n",
      "6/6 [==============================] - 0s 26ms/step - loss: 0.0012 - mae: 0.0299\n",
      "Epoch 91/400\n",
      "6/6 [==============================] - 0s 25ms/step - loss: 0.0012 - mae: 0.0305\n",
      "Epoch 92/400\n",
      "6/6 [==============================] - 0s 25ms/step - loss: 0.0012 - mae: 0.0301\n",
      "Epoch 93/400\n",
      "6/6 [==============================] - 0s 25ms/step - loss: 0.0012 - mae: 0.0310\n",
      "Epoch 94/400\n",
      "6/6 [==============================] - 0s 29ms/step - loss: 0.0012 - mae: 0.0300\n",
      "Epoch 95/400\n",
      "6/6 [==============================] - 0s 25ms/step - loss: 0.0012 - mae: 0.0305\n",
      "Epoch 96/400\n",
      "6/6 [==============================] - 0s 24ms/step - loss: 0.0012 - mae: 0.0300\n",
      "Epoch 97/400\n",
      "6/6 [==============================] - 0s 28ms/step - loss: 0.0012 - mae: 0.0312\n",
      "Epoch 98/400\n",
      "6/6 [==============================] - 0s 30ms/step - loss: 0.0012 - mae: 0.0295\n",
      "Epoch 99/400\n",
      "6/6 [==============================] - 0s 28ms/step - loss: 0.0012 - mae: 0.0301\n",
      "Epoch 100/400\n",
      "6/6 [==============================] - 0s 32ms/step - loss: 0.0012 - mae: 0.0295\n",
      "Epoch 101/400\n",
      "6/6 [==============================] - 0s 26ms/step - loss: 0.0012 - mae: 0.0299\n",
      "Epoch 102/400\n",
      "6/6 [==============================] - 0s 30ms/step - loss: 0.0012 - mae: 0.0295\n",
      "Epoch 103/400\n",
      "6/6 [==============================] - 0s 29ms/step - loss: 0.0012 - mae: 0.0299\n",
      "Epoch 104/400\n",
      "6/6 [==============================] - 0s 31ms/step - loss: 0.0012 - mae: 0.0296\n",
      "Epoch 105/400\n",
      "6/6 [==============================] - 0s 27ms/step - loss: 0.0012 - mae: 0.0299\n",
      "Epoch 106/400\n",
      "6/6 [==============================] - 0s 25ms/step - loss: 0.0012 - mae: 0.0295\n",
      "Epoch 107/400\n",
      "6/6 [==============================] - 0s 25ms/step - loss: 0.0012 - mae: 0.0310\n",
      "Epoch 108/400\n",
      "6/6 [==============================] - 0s 22ms/step - loss: 0.0012 - mae: 0.0298\n",
      "Epoch 109/400\n",
      "6/6 [==============================] - 0s 30ms/step - loss: 0.0012 - mae: 0.0303\n",
      "Epoch 110/400\n",
      "6/6 [==============================] - 0s 31ms/step - loss: 0.0012 - mae: 0.0298\n",
      "Epoch 111/400\n",
      "6/6 [==============================] - 0s 20ms/step - loss: 0.0012 - mae: 0.0295\n",
      "Epoch 112/400\n",
      "6/6 [==============================] - 0s 26ms/step - loss: 0.0012 - mae: 0.0295\n",
      "Epoch 113/400\n",
      "6/6 [==============================] - 0s 24ms/step - loss: 0.0012 - mae: 0.0301\n",
      "Epoch 114/400\n",
      "6/6 [==============================] - 0s 27ms/step - loss: 0.0012 - mae: 0.0294\n",
      "Epoch 115/400\n",
      "6/6 [==============================] - 0s 21ms/step - loss: 0.0012 - mae: 0.0299\n",
      "Epoch 116/400\n",
      "6/6 [==============================] - 0s 20ms/step - loss: 0.0012 - mae: 0.0293\n",
      "Epoch 117/400\n",
      "6/6 [==============================] - 0s 29ms/step - loss: 0.0012 - mae: 0.0295\n",
      "Epoch 118/400\n",
      "6/6 [==============================] - 0s 34ms/step - loss: 0.0012 - mae: 0.0292\n",
      "Epoch 119/400\n",
      "6/6 [==============================] - 0s 35ms/step - loss: 0.0012 - mae: 0.0299\n",
      "Epoch 120/400\n",
      "6/6 [==============================] - 0s 31ms/step - loss: 0.0012 - mae: 0.0293\n",
      "Epoch 121/400\n",
      "6/6 [==============================] - 0s 31ms/step - loss: 0.0012 - mae: 0.0297\n",
      "Epoch 122/400\n",
      "6/6 [==============================] - 0s 25ms/step - loss: 0.0012 - mae: 0.0293\n",
      "Epoch 123/400\n",
      "6/6 [==============================] - 0s 31ms/step - loss: 0.0012 - mae: 0.0294\n",
      "Epoch 124/400\n",
      "6/6 [==============================] - 0s 34ms/step - loss: 0.0012 - mae: 0.0292\n",
      "Epoch 125/400\n",
      "6/6 [==============================] - 0s 34ms/step - loss: 0.0012 - mae: 0.0294\n",
      "Epoch 126/400\n",
      "6/6 [==============================] - 0s 30ms/step - loss: 0.0012 - mae: 0.0291\n",
      "Epoch 127/400\n",
      "6/6 [==============================] - 0s 29ms/step - loss: 0.0012 - mae: 0.0298\n",
      "Epoch 128/400\n",
      "6/6 [==============================] - 0s 44ms/step - loss: 0.0012 - mae: 0.0292\n",
      "Epoch 129/400\n",
      "6/6 [==============================] - 0s 21ms/step - loss: 0.0012 - mae: 0.0302\n",
      "Epoch 130/400\n",
      "6/6 [==============================] - 0s 25ms/step - loss: 0.0012 - mae: 0.0292\n",
      "Epoch 131/400\n",
      "6/6 [==============================] - 0s 26ms/step - loss: 0.0012 - mae: 0.0306\n",
      "Epoch 132/400\n",
      "6/6 [==============================] - 0s 24ms/step - loss: 0.0012 - mae: 0.0293\n",
      "Epoch 133/400\n",
      "6/6 [==============================] - 0s 31ms/step - loss: 0.0012 - mae: 0.0300\n",
      "Epoch 134/400\n",
      "6/6 [==============================] - 0s 29ms/step - loss: 0.0012 - mae: 0.0290\n",
      "Epoch 135/400\n",
      "6/6 [==============================] - 0s 29ms/step - loss: 0.0012 - mae: 0.0292\n",
      "Epoch 136/400\n",
      "6/6 [==============================] - 0s 28ms/step - loss: 0.0012 - mae: 0.0298\n",
      "Epoch 137/400\n",
      "6/6 [==============================] - 0s 39ms/step - loss: 0.0012 - mae: 0.0291\n",
      "Epoch 138/400\n",
      "6/6 [==============================] - 0s 33ms/step - loss: 0.0012 - mae: 0.0299\n",
      "Epoch 139/400\n",
      "6/6 [==============================] - 0s 32ms/step - loss: 0.0012 - mae: 0.0293\n",
      "Epoch 140/400\n",
      "6/6 [==============================] - 0s 32ms/step - loss: 0.0012 - mae: 0.0306\n",
      "Epoch 141/400\n",
      "6/6 [==============================] - 0s 34ms/step - loss: 0.0012 - mae: 0.0291\n",
      "Epoch 142/400\n",
      "6/6 [==============================] - 0s 28ms/step - loss: 0.0012 - mae: 0.0303\n",
      "Epoch 143/400\n",
      "6/6 [==============================] - 0s 35ms/step - loss: 0.0012 - mae: 0.0290\n",
      "Epoch 144/400\n",
      "6/6 [==============================] - 0s 32ms/step - loss: 0.0012 - mae: 0.0300\n",
      "Epoch 145/400\n",
      "6/6 [==============================] - 0s 36ms/step - loss: 0.0012 - mae: 0.0293\n",
      "Epoch 146/400\n",
      "6/6 [==============================] - 0s 34ms/step - loss: 0.0012 - mae: 0.0293\n",
      "Epoch 147/400\n",
      "6/6 [==============================] - 0s 42ms/step - loss: 0.0012 - mae: 0.0293\n",
      "Epoch 148/400\n",
      "6/6 [==============================] - 0s 37ms/step - loss: 0.0012 - mae: 0.0290\n",
      "Epoch 149/400\n",
      "6/6 [==============================] - 0s 29ms/step - loss: 0.0012 - mae: 0.0296\n",
      "Epoch 150/400\n",
      "6/6 [==============================] - 0s 27ms/step - loss: 0.0011 - mae: 0.0289\n",
      "Epoch 151/400\n",
      "6/6 [==============================] - 0s 25ms/step - loss: 0.0011 - mae: 0.0291\n",
      "Epoch 152/400\n",
      "6/6 [==============================] - 0s 27ms/step - loss: 0.0011 - mae: 0.0288\n",
      "Epoch 153/400\n",
      "6/6 [==============================] - 0s 32ms/step - loss: 0.0012 - mae: 0.0289\n",
      "Epoch 154/400\n",
      "6/6 [==============================] - 0s 29ms/step - loss: 0.0012 - mae: 0.0287\n",
      "Epoch 155/400\n",
      "6/6 [==============================] - 0s 35ms/step - loss: 0.0012 - mae: 0.0300\n",
      "Epoch 156/400\n",
      "6/6 [==============================] - 0s 35ms/step - loss: 0.0012 - mae: 0.0289\n",
      "Epoch 157/400\n",
      "6/6 [==============================] - 0s 28ms/step - loss: 0.0012 - mae: 0.0298\n",
      "Epoch 158/400\n",
      "6/6 [==============================] - 0s 26ms/step - loss: 0.0012 - mae: 0.0290\n",
      "Epoch 159/400\n",
      "6/6 [==============================] - 0s 28ms/step - loss: 0.0011 - mae: 0.0292\n",
      "Epoch 160/400\n",
      "6/6 [==============================] - 0s 22ms/step - loss: 0.0011 - mae: 0.0290\n",
      "Epoch 161/400\n",
      "6/6 [==============================] - 0s 27ms/step - loss: 0.0011 - mae: 0.0288\n",
      "Epoch 162/400\n",
      "6/6 [==============================] - 0s 25ms/step - loss: 0.0012 - mae: 0.0288\n",
      "Epoch 163/400\n",
      "6/6 [==============================] - 0s 33ms/step - loss: 0.0011 - mae: 0.0289\n",
      "Epoch 164/400\n",
      "6/6 [==============================] - 0s 36ms/step - loss: 0.0011 - mae: 0.0288\n",
      "Epoch 165/400\n",
      "6/6 [==============================] - 0s 30ms/step - loss: 0.0011 - mae: 0.0291\n",
      "Epoch 166/400\n",
      "6/6 [==============================] - 0s 27ms/step - loss: 0.0011 - mae: 0.0288\n",
      "Epoch 167/400\n",
      "6/6 [==============================] - 0s 33ms/step - loss: 0.0011 - mae: 0.0289\n",
      "Epoch 168/400\n",
      "6/6 [==============================] - 0s 33ms/step - loss: 0.0011 - mae: 0.0287\n",
      "Epoch 169/400\n",
      "6/6 [==============================] - 0s 40ms/step - loss: 0.0012 - mae: 0.0287\n",
      "Epoch 170/400\n",
      "6/6 [==============================] - 0s 30ms/step - loss: 0.0011 - mae: 0.0288\n",
      "Epoch 171/400\n",
      "6/6 [==============================] - 0s 32ms/step - loss: 0.0011 - mae: 0.0291\n",
      "Epoch 172/400\n",
      "6/6 [==============================] - 0s 24ms/step - loss: 0.0011 - mae: 0.0287\n",
      "Epoch 173/400\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "6/6 [==============================] - 0s 37ms/step - loss: 0.0011 - mae: 0.0288\n",
      "Epoch 174/400\n",
      "6/6 [==============================] - 0s 40ms/step - loss: 0.0011 - mae: 0.0286\n",
      "Epoch 175/400\n",
      "6/6 [==============================] - 0s 38ms/step - loss: 0.0011 - mae: 0.0288\n",
      "Epoch 176/400\n",
      "6/6 [==============================] - 0s 26ms/step - loss: 0.0011 - mae: 0.0285\n",
      "Epoch 177/400\n",
      "6/6 [==============================] - 0s 38ms/step - loss: 0.0011 - mae: 0.0288\n",
      "Epoch 178/400\n",
      "6/6 [==============================] - 0s 43ms/step - loss: 0.0011 - mae: 0.0289\n",
      "Epoch 179/400\n",
      "6/6 [==============================] - 0s 23ms/step - loss: 0.0012 - mae: 0.0286\n",
      "Epoch 180/400\n",
      "6/6 [==============================] - 0s 28ms/step - loss: 0.0011 - mae: 0.0289\n",
      "Epoch 181/400\n",
      "6/6 [==============================] - 0s 26ms/step - loss: 0.0011 - mae: 0.0285\n",
      "Epoch 182/400\n",
      "6/6 [==============================] - 0s 32ms/step - loss: 0.0012 - mae: 0.0287\n",
      "Epoch 183/400\n",
      "6/6 [==============================] - 0s 24ms/step - loss: 0.0011 - mae: 0.0289\n",
      "Epoch 184/400\n",
      "6/6 [==============================] - 0s 30ms/step - loss: 0.0011 - mae: 0.0286\n",
      "Epoch 185/400\n",
      "6/6 [==============================] - 0s 32ms/step - loss: 0.0011 - mae: 0.0285\n",
      "Epoch 186/400\n",
      "6/6 [==============================] - 0s 30ms/step - loss: 0.0011 - mae: 0.0284\n",
      "Epoch 187/400\n",
      "6/6 [==============================] - 0s 35ms/step - loss: 0.0011 - mae: 0.0287\n",
      "Epoch 188/400\n",
      "6/6 [==============================] - 0s 30ms/step - loss: 0.0011 - mae: 0.0285\n",
      "Epoch 189/400\n",
      "6/6 [==============================] - 0s 27ms/step - loss: 0.0011 - mae: 0.0285\n",
      "Epoch 190/400\n",
      "6/6 [==============================] - 0s 31ms/step - loss: 0.0011 - mae: 0.0283\n",
      "Epoch 191/400\n",
      "6/6 [==============================] - 0s 24ms/step - loss: 0.0011 - mae: 0.0289\n",
      "Epoch 192/400\n",
      "6/6 [==============================] - 0s 32ms/step - loss: 0.0011 - mae: 0.0284\n",
      "Epoch 193/400\n",
      "6/6 [==============================] - 0s 31ms/step - loss: 0.0011 - mae: 0.0285\n",
      "Epoch 194/400\n",
      "6/6 [==============================] - 0s 29ms/step - loss: 0.0011 - mae: 0.0287\n",
      "Epoch 195/400\n",
      "6/6 [==============================] - 0s 25ms/step - loss: 0.0011 - mae: 0.0283\n",
      "Epoch 196/400\n",
      "6/6 [==============================] - 0s 26ms/step - loss: 0.0011 - mae: 0.0293\n",
      "Epoch 197/400\n",
      "6/6 [==============================] - 0s 27ms/step - loss: 0.0011 - mae: 0.0288\n",
      "Epoch 198/400\n",
      "6/6 [==============================] - 0s 34ms/step - loss: 0.0011 - mae: 0.0285\n",
      "Epoch 199/400\n",
      "6/6 [==============================] - 0s 25ms/step - loss: 0.0011 - mae: 0.0290\n",
      "Epoch 200/400\n",
      "6/6 [==============================] - 0s 21ms/step - loss: 0.0011 - mae: 0.0284\n",
      "Epoch 201/400\n",
      "6/6 [==============================] - 0s 25ms/step - loss: 0.0011 - mae: 0.0285\n",
      "Epoch 202/400\n",
      "6/6 [==============================] - 0s 28ms/step - loss: 0.0011 - mae: 0.0283\n",
      "Epoch 203/400\n",
      "6/6 [==============================] - 0s 25ms/step - loss: 0.0011 - mae: 0.0285\n",
      "Epoch 204/400\n",
      "6/6 [==============================] - 0s 21ms/step - loss: 0.0011 - mae: 0.0284\n",
      "Epoch 205/400\n",
      "6/6 [==============================] - 0s 21ms/step - loss: 0.0011 - mae: 0.0288\n",
      "Epoch 206/400\n",
      "6/6 [==============================] - 0s 20ms/step - loss: 0.0011 - mae: 0.0290\n",
      "Epoch 207/400\n",
      "6/6 [==============================] - 0s 23ms/step - loss: 0.0011 - mae: 0.0288\n",
      "Epoch 208/400\n",
      "6/6 [==============================] - ETA: 0s - loss: 9.3310e-04 - mae: 0.027 - 0s 23ms/step - loss: 0.0011 - mae: 0.0283\n",
      "Epoch 209/400\n",
      "6/6 [==============================] - 0s 22ms/step - loss: 0.0011 - mae: 0.0282\n",
      "Epoch 210/400\n",
      "6/6 [==============================] - 0s 23ms/step - loss: 0.0011 - mae: 0.0289\n",
      "Epoch 211/400\n",
      "6/6 [==============================] - 0s 23ms/step - loss: 0.0012 - mae: 0.0282\n",
      "Epoch 212/400\n",
      "6/6 [==============================] - 0s 25ms/step - loss: 0.0012 - mae: 0.0300\n",
      "Epoch 213/400\n",
      "6/6 [==============================] - 0s 24ms/step - loss: 0.0011 - mae: 0.0281\n",
      "Epoch 214/400\n",
      "6/6 [==============================] - 0s 23ms/step - loss: 0.0011 - mae: 0.0285\n",
      "Epoch 215/400\n",
      "6/6 [==============================] - 0s 23ms/step - loss: 0.0011 - mae: 0.0282\n",
      "Epoch 216/400\n",
      "6/6 [==============================] - 0s 20ms/step - loss: 0.0011 - mae: 0.0281\n",
      "Epoch 217/400\n",
      "6/6 [==============================] - 0s 19ms/step - loss: 0.0011 - mae: 0.0280\n",
      "Epoch 218/400\n",
      "6/6 [==============================] - 0s 23ms/step - loss: 0.0011 - mae: 0.0284\n",
      "Epoch 219/400\n",
      "6/6 [==============================] - 0s 23ms/step - loss: 0.0011 - mae: 0.0281\n",
      "Epoch 220/400\n",
      "6/6 [==============================] - 0s 22ms/step - loss: 0.0011 - mae: 0.0283\n",
      "Epoch 221/400\n",
      "6/6 [==============================] - 0s 19ms/step - loss: 0.0011 - mae: 0.0280\n",
      "Epoch 222/400\n",
      "6/6 [==============================] - 0s 18ms/step - loss: 0.0011 - mae: 0.0284\n",
      "Epoch 223/400\n",
      "6/6 [==============================] - 0s 17ms/step - loss: 0.0011 - mae: 0.0280\n",
      "Epoch 224/400\n",
      "6/6 [==============================] - 0s 25ms/step - loss: 0.0011 - mae: 0.0282\n",
      "Epoch 225/400\n",
      "6/6 [==============================] - 0s 29ms/step - loss: 0.0011 - mae: 0.0285\n",
      "Epoch 226/400\n",
      "6/6 [==============================] - 0s 26ms/step - loss: 0.0011 - mae: 0.0284\n",
      "Epoch 227/400\n",
      "6/6 [==============================] - 0s 25ms/step - loss: 0.0011 - mae: 0.0287\n",
      "Epoch 228/400\n",
      "6/6 [==============================] - 0s 27ms/step - loss: 0.0011 - mae: 0.0292\n",
      "Epoch 229/400\n",
      "6/6 [==============================] - 0s 28ms/step - loss: 0.0011 - mae: 0.0284\n",
      "Epoch 230/400\n",
      "6/6 [==============================] - 0s 27ms/step - loss: 0.0011 - mae: 0.0286\n",
      "Epoch 231/400\n",
      "6/6 [==============================] - 0s 28ms/step - loss: 0.0011 - mae: 0.0279\n",
      "Epoch 232/400\n",
      "6/6 [==============================] - 0s 25ms/step - loss: 0.0011 - mae: 0.0301\n",
      "Epoch 233/400\n",
      "6/6 [==============================] - 0s 26ms/step - loss: 0.0011 - mae: 0.0284\n",
      "Epoch 234/400\n",
      "6/6 [==============================] - 0s 26ms/step - loss: 0.0011 - mae: 0.0286\n",
      "Epoch 235/400\n",
      "6/6 [==============================] - 0s 24ms/step - loss: 0.0011 - mae: 0.0279\n",
      "Epoch 236/400\n",
      "6/6 [==============================] - 0s 31ms/step - loss: 0.0011 - mae: 0.0289\n",
      "Epoch 237/400\n",
      "6/6 [==============================] - 0s 26ms/step - loss: 0.0011 - mae: 0.0282\n",
      "Epoch 238/400\n",
      "6/6 [==============================] - 0s 23ms/step - loss: 0.0011 - mae: 0.0292\n",
      "Epoch 239/400\n",
      "6/6 [==============================] - 0s 25ms/step - loss: 0.0011 - mae: 0.0281\n",
      "Epoch 240/400\n",
      "6/6 [==============================] - 0s 33ms/step - loss: 0.0011 - mae: 0.0293\n",
      "Epoch 241/400\n",
      "6/6 [==============================] - 0s 24ms/step - loss: 0.0011 - mae: 0.0283\n",
      "Epoch 242/400\n",
      "6/6 [==============================] - 0s 22ms/step - loss: 0.0011 - mae: 0.0288\n",
      "Epoch 243/400\n",
      "6/6 [==============================] - 0s 20ms/step - loss: 0.0011 - mae: 0.0279\n",
      "Epoch 244/400\n",
      "6/6 [==============================] - 0s 31ms/step - loss: 0.0011 - mae: 0.0289\n",
      "Epoch 245/400\n",
      "6/6 [==============================] - 0s 25ms/step - loss: 0.0011 - mae: 0.0281\n",
      "Epoch 246/400\n",
      "6/6 [==============================] - 0s 30ms/step - loss: 0.0011 - mae: 0.0279\n",
      "Epoch 247/400\n",
      "6/6 [==============================] - 0s 26ms/step - loss: 0.0011 - mae: 0.0280\n",
      "Epoch 248/400\n",
      "6/6 [==============================] - 0s 22ms/step - loss: 0.0011 - mae: 0.0278\n",
      "Epoch 249/400\n",
      "6/6 [==============================] - 0s 28ms/step - loss: 0.0011 - mae: 0.0279\n",
      "Epoch 250/400\n",
      "6/6 [==============================] - 0s 23ms/step - loss: 0.0011 - mae: 0.0279\n",
      "Epoch 251/400\n",
      "6/6 [==============================] - 0s 26ms/step - loss: 0.0011 - mae: 0.0276\n",
      "Epoch 252/400\n",
      "6/6 [==============================] - 0s 25ms/step - loss: 0.0011 - mae: 0.0283\n",
      "Epoch 253/400\n",
      "6/6 [==============================] - 0s 22ms/step - loss: 0.0011 - mae: 0.0277\n",
      "Epoch 254/400\n",
      "6/6 [==============================] - 0s 27ms/step - loss: 0.0011 - mae: 0.0279\n",
      "Epoch 255/400\n",
      "6/6 [==============================] - 0s 29ms/step - loss: 0.0011 - mae: 0.0278\n",
      "Epoch 256/400\n",
      "6/6 [==============================] - 0s 30ms/step - loss: 0.0011 - mae: 0.0278\n",
      "Epoch 257/400\n",
      "6/6 [==============================] - 0s 30ms/step - loss: 0.0011 - mae: 0.0282\n",
      "Epoch 258/400\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "6/6 [==============================] - 0s 23ms/step - loss: 0.0011 - mae: 0.0279\n",
      "Epoch 259/400\n",
      "6/6 [==============================] - 0s 28ms/step - loss: 0.0011 - mae: 0.0283\n",
      "Epoch 260/400\n",
      "6/6 [==============================] - 0s 30ms/step - loss: 0.0011 - mae: 0.0278\n",
      "Epoch 261/400\n",
      "6/6 [==============================] - 0s 36ms/step - loss: 0.0011 - mae: 0.0293\n",
      "Epoch 262/400\n",
      "6/6 [==============================] - 0s 29ms/step - loss: 0.0011 - mae: 0.0281\n",
      "Epoch 263/400\n",
      "6/6 [==============================] - 0s 23ms/step - loss: 0.0011 - mae: 0.0291\n",
      "Epoch 264/400\n",
      "6/6 [==============================] - 0s 22ms/step - loss: 0.0011 - mae: 0.0281\n",
      "Epoch 265/400\n",
      "6/6 [==============================] - 0s 26ms/step - loss: 0.0011 - mae: 0.0287\n",
      "Epoch 266/400\n",
      "6/6 [==============================] - 0s 20ms/step - loss: 0.0011 - mae: 0.0280\n",
      "Epoch 267/400\n",
      "6/6 [==============================] - 0s 26ms/step - loss: 0.0011 - mae: 0.0276\n",
      "Epoch 268/400\n",
      "6/6 [==============================] - 0s 27ms/step - loss: 0.0011 - mae: 0.0276\n",
      "Epoch 269/400\n",
      "6/6 [==============================] - 0s 23ms/step - loss: 0.0011 - mae: 0.0283\n",
      "Epoch 270/400\n",
      "6/6 [==============================] - 0s 23ms/step - loss: 0.0011 - mae: 0.0281\n",
      "Epoch 271/400\n",
      "6/6 [==============================] - 0s 21ms/step - loss: 0.0011 - mae: 0.0283\n",
      "Epoch 272/400\n",
      "6/6 [==============================] - 0s 23ms/step - loss: 0.0011 - mae: 0.0278\n",
      "Epoch 273/400\n",
      "6/6 [==============================] - 0s 29ms/step - loss: 0.0011 - mae: 0.0276\n",
      "Epoch 274/400\n",
      "6/6 [==============================] - 0s 25ms/step - loss: 0.0011 - mae: 0.0278\n",
      "Epoch 275/400\n",
      "6/6 [==============================] - 0s 21ms/step - loss: 0.0011 - mae: 0.0276\n",
      "Epoch 276/400\n",
      "6/6 [==============================] - 0s 23ms/step - loss: 0.0011 - mae: 0.0284\n",
      "Epoch 277/400\n",
      "6/6 [==============================] - 0s 26ms/step - loss: 0.0011 - mae: 0.0278\n",
      "Epoch 278/400\n",
      "6/6 [==============================] - 0s 28ms/step - loss: 0.0011 - mae: 0.0293\n",
      "Epoch 279/400\n",
      "6/6 [==============================] - 0s 26ms/step - loss: 0.0011 - mae: 0.0283\n",
      "Epoch 280/400\n",
      "6/6 [==============================] - 0s 27ms/step - loss: 0.0011 - mae: 0.0289\n",
      "Epoch 281/400\n",
      "6/6 [==============================] - 0s 33ms/step - loss: 0.0011 - mae: 0.0275\n",
      "Epoch 282/400\n",
      "6/6 [==============================] - 0s 33ms/step - loss: 0.0011 - mae: 0.0291\n",
      "Epoch 283/400\n",
      "6/6 [==============================] - 0s 25ms/step - loss: 0.0011 - mae: 0.0285\n",
      "Epoch 284/400\n",
      "6/6 [==============================] - 0s 21ms/step - loss: 0.0011 - mae: 0.0284\n",
      "Epoch 285/400\n",
      "6/6 [==============================] - 0s 21ms/step - loss: 0.0011 - mae: 0.0277\n",
      "Epoch 286/400\n",
      "6/6 [==============================] - 0s 26ms/step - loss: 0.0011 - mae: 0.0275\n",
      "Epoch 287/400\n",
      "6/6 [==============================] - 0s 24ms/step - loss: 0.0011 - mae: 0.0274\n",
      "Epoch 288/400\n",
      "6/6 [==============================] - 0s 22ms/step - loss: 0.0011 - mae: 0.0281\n",
      "Epoch 289/400\n",
      "6/6 [==============================] - 0s 22ms/step - loss: 0.0011 - mae: 0.0274\n",
      "Epoch 290/400\n",
      "6/6 [==============================] - 0s 23ms/step - loss: 0.0011 - mae: 0.0290\n",
      "Epoch 291/400\n",
      "6/6 [==============================] - 0s 21ms/step - loss: 0.0011 - mae: 0.0280\n",
      "Epoch 292/400\n",
      "6/6 [==============================] - 0s 24ms/step - loss: 0.0011 - mae: 0.0284\n",
      "Epoch 293/400\n",
      "6/6 [==============================] - 0s 27ms/step - loss: 0.0011 - mae: 0.0276\n",
      "Epoch 294/400\n",
      "6/6 [==============================] - 0s 19ms/step - loss: 0.0011 - mae: 0.0290\n",
      "Epoch 295/400\n",
      "6/6 [==============================] - 0s 26ms/step - loss: 0.0011 - mae: 0.0277\n",
      "Epoch 296/400\n",
      "6/6 [==============================] - 0s 27ms/step - loss: 0.0011 - mae: 0.0280\n",
      "Epoch 297/400\n",
      "6/6 [==============================] - 0s 28ms/step - loss: 0.0010 - mae: 0.0273\n",
      "Epoch 298/400\n",
      "6/6 [==============================] - 0s 28ms/step - loss: 0.0011 - mae: 0.0276\n",
      "Epoch 299/400\n",
      "6/6 [==============================] - 0s 24ms/step - loss: 0.0011 - mae: 0.0274\n",
      "Epoch 300/400\n",
      "6/6 [==============================] - 0s 29ms/step - loss: 0.0011 - mae: 0.0280\n",
      "Epoch 301/400\n",
      "6/6 [==============================] - 0s 35ms/step - loss: 0.0011 - mae: 0.0275\n",
      "Epoch 302/400\n",
      "6/6 [==============================] - 0s 33ms/step - loss: 0.0011 - mae: 0.0277\n",
      "Epoch 303/400\n",
      "6/6 [==============================] - 0s 24ms/step - loss: 0.0010 - mae: 0.0276\n",
      "Epoch 304/400\n",
      "6/6 [==============================] - 0s 26ms/step - loss: 0.0011 - mae: 0.0275\n",
      "Epoch 305/400\n",
      "6/6 [==============================] - 0s 23ms/step - loss: 0.0011 - mae: 0.0281\n",
      "Epoch 306/400\n",
      "6/6 [==============================] - 0s 24ms/step - loss: 0.0011 - mae: 0.0277\n",
      "Epoch 307/400\n",
      "6/6 [==============================] - 0s 26ms/step - loss: 0.0011 - mae: 0.0274\n",
      "Epoch 308/400\n",
      "6/6 [==============================] - 0s 24ms/step - loss: 0.0010 - mae: 0.0275\n",
      "Epoch 309/400\n",
      "6/6 [==============================] - 0s 25ms/step - loss: 0.0010 - mae: 0.0273\n",
      "Epoch 310/400\n",
      "6/6 [==============================] - 0s 27ms/step - loss: 0.0011 - mae: 0.0274\n",
      "Epoch 311/400\n",
      "6/6 [==============================] - 0s 33ms/step - loss: 0.0010 - mae: 0.0275\n",
      "Epoch 312/400\n",
      "6/6 [==============================] - 0s 29ms/step - loss: 0.0011 - mae: 0.0273\n",
      "Epoch 313/400\n",
      "6/6 [==============================] - 0s 23ms/step - loss: 0.0010 - mae: 0.0278\n",
      "Epoch 314/400\n",
      "6/6 [==============================] - 0s 26ms/step - loss: 0.0011 - mae: 0.0274\n",
      "Epoch 315/400\n",
      "6/6 [==============================] - 0s 29ms/step - loss: 0.0010 - mae: 0.0281\n",
      "Epoch 316/400\n",
      "6/6 [==============================] - 0s 34ms/step - loss: 0.0011 - mae: 0.0275\n",
      "Epoch 317/400\n",
      "6/6 [==============================] - 0s 28ms/step - loss: 0.0011 - mae: 0.0282\n",
      "Epoch 318/400\n",
      "6/6 [==============================] - 0s 26ms/step - loss: 0.0011 - mae: 0.0277\n",
      "Epoch 319/400\n",
      "6/6 [==============================] - 0s 28ms/step - loss: 0.0011 - mae: 0.0284\n",
      "Epoch 320/400\n",
      "6/6 [==============================] - 0s 23ms/step - loss: 0.0010 - mae: 0.0276\n",
      "Epoch 321/400\n",
      "6/6 [==============================] - 0s 25ms/step - loss: 0.0011 - mae: 0.0277\n",
      "Epoch 322/400\n",
      "6/6 [==============================] - 0s 31ms/step - loss: 0.0011 - mae: 0.0278\n",
      "Epoch 323/400\n",
      "6/6 [==============================] - 0s 41ms/step - loss: 0.0010 - mae: 0.0272\n",
      "Epoch 324/400\n",
      "6/6 [==============================] - 0s 30ms/step - loss: 0.0010 - mae: 0.0276\n",
      "Epoch 325/400\n",
      "6/6 [==============================] - 0s 27ms/step - loss: 0.0010 - mae: 0.0271\n",
      "Epoch 326/400\n",
      "6/6 [==============================] - 0s 25ms/step - loss: 0.0010 - mae: 0.0276\n",
      "Epoch 327/400\n",
      "6/6 [==============================] - 0s 27ms/step - loss: 0.0010 - mae: 0.0272\n",
      "Epoch 328/400\n",
      "6/6 [==============================] - 0s 37ms/step - loss: 0.0011 - mae: 0.0276\n",
      "Epoch 329/400\n",
      "6/6 [==============================] - 0s 32ms/step - loss: 0.0010 - mae: 0.0275\n",
      "Epoch 330/400\n",
      "6/6 [==============================] - 0s 28ms/step - loss: 0.0010 - mae: 0.0272\n",
      "Epoch 331/400\n",
      "6/6 [==============================] - 0s 44ms/step - loss: 0.0010 - mae: 0.0278\n",
      "Epoch 332/400\n",
      "6/6 [==============================] - 0s 39ms/step - loss: 0.0010 - mae: 0.0272\n",
      "Epoch 333/400\n",
      "6/6 [==============================] - 0s 44ms/step - loss: 0.0011 - mae: 0.0281\n",
      "Epoch 334/400\n",
      "6/6 [==============================] - 0s 34ms/step - loss: 0.0011 - mae: 0.0275\n",
      "Epoch 335/400\n",
      "6/6 [==============================] - 0s 28ms/step - loss: 0.0010 - mae: 0.0280\n",
      "Epoch 336/400\n",
      "6/6 [==============================] - 0s 32ms/step - loss: 0.0011 - mae: 0.0273\n",
      "Epoch 337/400\n",
      "6/6 [==============================] - 0s 30ms/step - loss: 0.0010 - mae: 0.0274\n",
      "Epoch 338/400\n",
      "6/6 [==============================] - 0s 32ms/step - loss: 0.0010 - mae: 0.0270\n",
      "Epoch 339/400\n",
      "6/6 [==============================] - 0s 50ms/step - loss: 0.0011 - mae: 0.0275\n",
      "Epoch 340/400\n",
      "6/6 [==============================] - 0s 28ms/step - loss: 0.0010 - mae: 0.0278\n",
      "Epoch 341/400\n",
      "6/6 [==============================] - 0s 28ms/step - loss: 0.0010 - mae: 0.0273\n",
      "Epoch 342/400\n",
      "6/6 [==============================] - 0s 24ms/step - loss: 0.0011 - mae: 0.0287\n",
      "Epoch 343/400\n",
      "6/6 [==============================] - 0s 23ms/step - loss: 0.0011 - mae: 0.0271\n",
      "Epoch 344/400\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "6/6 [==============================] - 0s 26ms/step - loss: 0.0011 - mae: 0.0290\n",
      "Epoch 345/400\n",
      "6/6 [==============================] - 0s 22ms/step - loss: 0.0011 - mae: 0.0281\n",
      "Epoch 346/400\n",
      "6/6 [==============================] - 0s 24ms/step - loss: 0.0010 - mae: 0.0274\n",
      "Epoch 347/400\n",
      "6/6 [==============================] - 0s 27ms/step - loss: 0.0010 - mae: 0.0271\n",
      "Epoch 348/400\n",
      "6/6 [==============================] - 0s 23ms/step - loss: 0.0011 - mae: 0.0282\n",
      "Epoch 349/400\n",
      "6/6 [==============================] - 0s 26ms/step - loss: 0.0010 - mae: 0.0278\n",
      "Epoch 350/400\n",
      "6/6 [==============================] - 0s 25ms/step - loss: 0.0011 - mae: 0.0271\n",
      "Epoch 351/400\n",
      "6/6 [==============================] - 0s 27ms/step - loss: 0.0011 - mae: 0.0286\n",
      "Epoch 352/400\n",
      "6/6 [==============================] - 0s 30ms/step - loss: 0.0010 - mae: 0.0274\n",
      "Epoch 353/400\n",
      "6/6 [==============================] - 0s 34ms/step - loss: 0.0011 - mae: 0.0283\n",
      "Epoch 354/400\n",
      "6/6 [==============================] - 0s 39ms/step - loss: 0.0011 - mae: 0.0275\n",
      "Epoch 355/400\n",
      "6/6 [==============================] - 0s 30ms/step - loss: 0.0010 - mae: 0.0274\n",
      "Epoch 356/400\n",
      "6/6 [==============================] - 0s 38ms/step - loss: 0.0010 - mae: 0.0273\n",
      "Epoch 357/400\n",
      "6/6 [==============================] - 0s 30ms/step - loss: 0.0010 - mae: 0.0273\n",
      "Epoch 358/400\n",
      "6/6 [==============================] - 0s 34ms/step - loss: 0.0010 - mae: 0.0271\n",
      "Epoch 359/400\n",
      "6/6 [==============================] - 0s 34ms/step - loss: 0.0011 - mae: 0.0281\n",
      "Epoch 360/400\n",
      "6/6 [==============================] - 0s 28ms/step - loss: 0.0011 - mae: 0.0279\n",
      "Epoch 361/400\n",
      "6/6 [==============================] - 0s 27ms/step - loss: 0.0010 - mae: 0.0272\n",
      "Epoch 362/400\n",
      "6/6 [==============================] - 0s 30ms/step - loss: 0.0010 - mae: 0.0272\n",
      "Epoch 363/400\n",
      "6/6 [==============================] - 0s 30ms/step - loss: 0.0010 - mae: 0.0272\n",
      "Epoch 364/400\n",
      "6/6 [==============================] - 0s 39ms/step - loss: 0.0010 - mae: 0.0271\n",
      "Epoch 365/400\n",
      "6/6 [==============================] - 0s 32ms/step - loss: 0.0010 - mae: 0.0272\n",
      "Epoch 366/400\n",
      "6/6 [==============================] - 0s 30ms/step - loss: 0.0011 - mae: 0.0273\n",
      "Epoch 367/400\n",
      "6/6 [==============================] - 0s 23ms/step - loss: 0.0010 - mae: 0.0271\n",
      "Epoch 368/400\n",
      "6/6 [==============================] - 0s 36ms/step - loss: 0.0011 - mae: 0.0271\n",
      "Epoch 369/400\n",
      "6/6 [==============================] - 0s 26ms/step - loss: 0.0010 - mae: 0.0273\n",
      "Epoch 370/400\n",
      "6/6 [==============================] - 0s 29ms/step - loss: 0.0011 - mae: 0.0270\n",
      "Epoch 371/400\n",
      "6/6 [==============================] - 0s 24ms/step - loss: 0.0010 - mae: 0.0279\n",
      "Epoch 372/400\n",
      "6/6 [==============================] - 0s 27ms/step - loss: 0.0010 - mae: 0.0270\n",
      "Epoch 373/400\n",
      "6/6 [==============================] - 0s 26ms/step - loss: 0.0010 - mae: 0.0278\n",
      "Epoch 374/400\n",
      "6/6 [==============================] - 0s 22ms/step - loss: 0.0010 - mae: 0.0273\n",
      "Epoch 375/400\n",
      "6/6 [==============================] - 0s 22ms/step - loss: 0.0010 - mae: 0.0270\n",
      "Epoch 376/400\n",
      "6/6 [==============================] - 0s 22ms/step - loss: 0.0010 - mae: 0.0275\n",
      "Epoch 377/400\n",
      "6/6 [==============================] - 0s 20ms/step - loss: 0.0010 - mae: 0.0271\n",
      "Epoch 378/400\n",
      "6/6 [==============================] - 0s 28ms/step - loss: 0.0010 - mae: 0.0275\n",
      "Epoch 379/400\n",
      "6/6 [==============================] - 0s 23ms/step - loss: 0.0010 - mae: 0.0273\n",
      "Epoch 380/400\n",
      "6/6 [==============================] - 0s 23ms/step - loss: 0.0011 - mae: 0.0272\n",
      "Epoch 381/400\n",
      "6/6 [==============================] - 0s 16ms/step - loss: 0.0011 - mae: 0.0286\n",
      "Epoch 382/400\n",
      "6/6 [==============================] - 0s 18ms/step - loss: 0.0010 - mae: 0.0274\n",
      "Epoch 383/400\n",
      "6/6 [==============================] - 0s 17ms/step - loss: 0.0010 - mae: 0.0273\n",
      "Epoch 384/400\n",
      "6/6 [==============================] - 0s 19ms/step - loss: 0.0011 - mae: 0.0279\n",
      "Epoch 385/400\n",
      "6/6 [==============================] - 0s 22ms/step - loss: 0.0010 - mae: 0.0273\n",
      "Epoch 386/400\n",
      "6/6 [==============================] - 0s 25ms/step - loss: 0.0010 - mae: 0.0278\n",
      "Epoch 387/400\n",
      "6/6 [==============================] - 0s 16ms/step - loss: 0.0010 - mae: 0.0271\n",
      "Epoch 388/400\n",
      "6/6 [==============================] - 0s 16ms/step - loss: 0.0010 - mae: 0.0270\n",
      "Epoch 389/400\n",
      "6/6 [==============================] - 0s 16ms/step - loss: 0.0010 - mae: 0.0268\n",
      "Epoch 390/400\n",
      "6/6 [==============================] - 0s 15ms/step - loss: 0.0010 - mae: 0.0280\n",
      "Epoch 391/400\n",
      "6/6 [==============================] - 0s 24ms/step - loss: 0.0010 - mae: 0.0272\n",
      "Epoch 392/400\n",
      "6/6 [==============================] - 0s 22ms/step - loss: 0.0010 - mae: 0.0272\n",
      "Epoch 393/400\n",
      "6/6 [==============================] - 0s 17ms/step - loss: 0.0010 - mae: 0.0277\n",
      "Epoch 394/400\n",
      "6/6 [==============================] - 0s 16ms/step - loss: 0.0010 - mae: 0.0268\n",
      "Epoch 395/400\n",
      "6/6 [==============================] - 0s 19ms/step - loss: 0.0010 - mae: 0.0270\n",
      "Epoch 396/400\n",
      "6/6 [==============================] - 0s 19ms/step - loss: 9.9912e-04 - mae: 0.0268\n",
      "Epoch 397/400\n",
      "6/6 [==============================] - 0s 18ms/step - loss: 0.0010 - mae: 0.0268\n",
      "Epoch 398/400\n",
      "6/6 [==============================] - 0s 16ms/step - loss: 0.0010 - mae: 0.0270\n",
      "Epoch 399/400\n",
      "6/6 [==============================] - 0s 23ms/step - loss: 0.0010 - mae: 0.0269\n",
      "Epoch 400/400\n",
      "6/6 [==============================] - 0s 23ms/step - loss: 0.0010 - mae: 0.0273\n"
     ]
    }
   ],
   "source": [
    "tf.keras.backend.clear_session()\n",
    "tf.random.set_seed(0)\n",
    "np.random.seed(0)\n",
    "\n",
    "window_size = 16\n",
    "batch_size = 128\n",
    "shuffle_buffer_size = 2000\n",
    "train_set = windowed_dataset(x_train, window_size, batch_size, shuffle_buffer_size)\n",
    "\n",
    "model = tf.keras.models.Sequential([\n",
    "  tf.keras.layers.Conv1D(filters=32, kernel_size=5,\n",
    "                      strides=1, padding=\"same\",\n",
    "                      activation=\"relu\",\n",
    "                      input_shape=[None, 1]),\n",
    "  tf.keras.layers.Bidirectional(tf.keras.layers.LSTM(64, return_sequences=True)),\n",
    "  tf.keras.layers.Bidirectional(tf.keras.layers.LSTM(64, return_sequences=True)),\n",
    "  tf.keras.layers.Dense(64),\n",
    "  tf.keras.layers.Dense(32),\n",
    "  tf.keras.layers.Dense(1, activation=\"tanh\"),\n",
    "  tf.keras.layers.Lambda(lambda x: x * 2)\n",
    "])\n",
    "\n",
    "optimizer = tf.keras.optimizers.Adam(learning_rate=5.e-5)\n",
    "model.compile(loss=tf.keras.losses.Huber(),\n",
    "              optimizer=optimizer,\n",
    "              metrics=[\"mae\"])\n",
    "\n",
    "history = model.fit(train_set, epochs=400)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x7f13f4449a20>]"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(np.arange(len(history.history[\"loss\"])), history.history[\"loss\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "MAE : 0.0329529270529747\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "rnn_forecast = model_forecast(model, series[..., np.newaxis], window_size)\n",
    "rnn_forecast = rnn_forecast[split_time - window_size:-1, -1, 0]\n",
    "\n",
    "plt.figure(figsize=(10, 6))\n",
    "plt.plot(time_valid, unnormalize_series(x_valid, mean_series, max_series), \"k\", linewidth=1, label=\"Validation data\")\n",
    "plt.plot(time_valid, unnormalize_series(rnn_forecast, mean_series, max_series), \"r\", linewidth=2, label=\"Model prediction\")\n",
    "plt.xlabel(\"Days\")\n",
    "plt.ylabel(\"Visitors\")\n",
    "plt.legend()\n",
    "\n",
    "print(\"MAE : {}\".format(tf.keras.metrics.mean_absolute_error(x_valid, rnn_forecast).numpy()))"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}