07_RNN_advanced_usage_of_rnn.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 기온 예측 문제"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "\n",
    "data_dir = \"datasets/jena_climate/\"\n",
    "fname = os.path.join(data_dir, 'jena_climate_2009_2016.csv')\n",
    "\n",
    "f = open(fname)\n",
    "data = f.read()\n",
    "f.close()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['\"Date Time\"', '\"p (mbar)\"', '\"T (degC)\"', '\"Tpot (K)\"', '\"Tdew (degC)\"', '\"rh (%)\"', '\"VPmax (mbar)\"', '\"VPact (mbar)\"', '\"VPdef (mbar)\"', '\"sh (g/kg)\"', '\"H2OC (mmol/mol)\"', '\"rho (g/m**3)\"', '\"wv (m/s)\"', '\"max. wv (m/s)\"', '\"wd (deg)\"']\n",
      "420551\n"
     ]
    }
   ],
   "source": [
    "lines = data.split('\\n')\n",
    "header = lines[0].split(',')\n",
    "lines = lines[1:]\n",
    "\n",
    "print(header)\n",
    "print(len(lines))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "\n",
    "float_data = np.zeros((len(lines), len(header) - 1))\n",
    "for i, line in enumerate(lines):\n",
    "    values = [float(x) for x in line.split(',')[1:]]\n",
    "    float_data[i, :] = values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x1c38ad19898>]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from matplotlib import pyplot as plt\n",
    "\n",
    "temp = float_data[:, 1]\n",
    "plt.plot(range(len(temp)), temp)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x1c38b512a90>]"
      ]
     },
     "execution_count": 5,
     "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(range(1440), temp[:1440])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "mean = float_data[:200000].mean(axis=0)\n",
    "float_data -= mean\n",
    "std = float_data[:200000].std(axis=0)\n",
    "float_data /= std"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "def generator(data, lookback, delay, min_index, max_index,\n",
    "              shuffle=False, batch_size=128, step=6):\n",
    "    if max_index is None:\n",
    "        max_index = len(data) - delay - 1\n",
    "    i = min_index + lookback\n",
    "    while True:\n",
    "        if shuffle:\n",
    "            rows = np.random.randint(\n",
    "                min_index + lookback, max_index, size =batch_size)\n",
    "        else:\n",
    "            if i + batch_size >= max_index:\n",
    "                i = min_index + lookback\n",
    "            rows = np.arange(i, min(i + batch_size, max_index))\n",
    "            i += len(rows)\n",
    "            \n",
    "        samples = np.zeros((len(rows),\n",
    "                            lookback // step,\n",
    "                            data.shape[-1]))\n",
    "        targets = np.zeros((len(rows),))\n",
    "        for j, row in enumerate(rows):\n",
    "            indices = range(rows[j] - lookback, rows[j], step)\n",
    "            samples[j] = data[indices]\n",
    "            targets[j] = data[rows[j] + delay][1]\n",
    "        yield samples, targets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "lookback = 1440\n",
    "step = 6\n",
    "delay = 144\n",
    "batch_size = 128\n",
    "\n",
    "train_gen = generator(float_data,\n",
    "                      lookback=lookback,\n",
    "                      delay=delay,\n",
    "                      min_index=0,\n",
    "                      max_index=200000,\n",
    "                      shuffle = True,\n",
    "                      step = step,\n",
    "                      batch_size =batch_size)\n",
    "val_gen = generator(float_data,\n",
    "                      lookback=lookback,\n",
    "                      delay=delay,\n",
    "                      min_index=200001,\n",
    "                      max_index=300000,\n",
    "                      step = step,\n",
    "                      batch_size =batch_size)\n",
    "test_gen = generator(float_data,\n",
    "                      lookback=lookback,\n",
    "                      delay=delay,\n",
    "                      min_index=300001,\n",
    "                      max_index=None,\n",
    "                      step = step,\n",
    "                      batch_size =batch_size)\n",
    "\n",
    "val_steps = (300000 - 200001 - lookback) // batch_size\n",
    "\n",
    "test_steps = (len(float_data) - 300001 - lookback) // batch_size"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.2897359729905486\n"
     ]
    }
   ],
   "source": [
    "def evaluate_native_method():\n",
    "    batch_maes = []\n",
    "    for step in range(val_steps):\n",
    "        samples, targets = next(val_gen)\n",
    "        preds = samples[:, -1, 1]\n",
    "        mae = np.mean(np.abs(preds - targets))\n",
    "        batch_maes.append(mae)\n",
    "    print(np.mean(batch_maes))\n",
    "    \n",
    "evaluate_native_method()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2.5672247338393395"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "celsius_mae = 0.29 * std[1]\n",
    "celsius_mae"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 기본적인 머신 러닝"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "from keras.models import Sequential\n",
    "from keras import layers\n",
    "from keras.optimizers import RMSprop"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/20\n",
      "500/500 [==============================] - 14s 29ms/step - loss: 1.4326 - val_loss: 0.6737\n",
      "Epoch 2/20\n",
      "500/500 [==============================] - 11s 22ms/step - loss: 0.4680 - val_loss: 0.4874\n",
      "Epoch 3/20\n",
      "500/500 [==============================] - 11s 22ms/step - loss: 0.2993 - val_loss: 0.3444\n",
      "Epoch 4/20\n",
      "500/500 [==============================] - 11s 22ms/step - loss: 0.2668 - val_loss: 0.3194\n",
      "Epoch 5/20\n",
      "500/500 [==============================] - 11s 22ms/step - loss: 0.2520 - val_loss: 0.3059\n",
      "Epoch 6/20\n",
      "500/500 [==============================] - 11s 22ms/step - loss: 0.2434 - val_loss: 0.3213\n",
      "Epoch 7/20\n",
      "500/500 [==============================] - 11s 22ms/step - loss: 0.2359 - val_loss: 0.3233\n",
      "Epoch 8/20\n",
      "500/500 [==============================] - 11s 23ms/step - loss: 0.2293 - val_loss: 0.3337\n",
      "Epoch 9/20\n",
      "500/500 [==============================] - 11s 23ms/step - loss: 0.2258 - val_loss: 0.3277\n",
      "Epoch 10/20\n",
      "500/500 [==============================] - 11s 22ms/step - loss: 0.2199 - val_loss: 0.3367\n",
      "Epoch 11/20\n",
      "500/500 [==============================] - 11s 22ms/step - loss: 0.2164 - val_loss: 0.3366\n",
      "Epoch 12/20\n",
      "500/500 [==============================] - 11s 22ms/step - loss: 0.2142 - val_loss: 0.3447\n",
      "Epoch 13/20\n",
      "500/500 [==============================] - 11s 22ms/step - loss: 0.2121 - val_loss: 0.3336\n",
      "Epoch 14/20\n",
      "500/500 [==============================] - 11s 22ms/step - loss: 0.2092 - val_loss: 0.3806\n",
      "Epoch 15/20\n",
      "500/500 [==============================] - 11s 22ms/step - loss: 0.2067 - val_loss: 0.3547\n",
      "Epoch 16/20\n",
      "500/500 [==============================] - 11s 23ms/step - loss: 0.2064 - val_loss: 0.3624\n",
      "Epoch 17/20\n",
      "500/500 [==============================] - 11s 22ms/step - loss: 0.2032 - val_loss: 0.3697\n",
      "Epoch 18/20\n",
      "500/500 [==============================] - 13s 25ms/step - loss: 0.2001 - val_loss: 0.3367\n",
      "Epoch 19/20\n",
      "500/500 [==============================] - 11s 23ms/step - loss: 0.1992 - val_loss: 0.3405\n",
      "Epoch 20/20\n",
      "500/500 [==============================] - 11s 23ms/step - loss: 0.1993 - val_loss: 0.3550\n"
     ]
    }
   ],
   "source": [
    "model = Sequential()\n",
    "model.add(layers.Flatten(input_shape=(lookback // step, float_data.shape[-1])))\n",
    "model.add(layers.Dense(32, activation='relu'))\n",
    "model.add(layers.Dense(1))\n",
    "\n",
    "model.compile(optimizer=RMSprop(), loss='mae')\n",
    "history = model.fit_generator(train_gen,\n",
    "                              steps_per_epoch=500,\n",
    "                              epochs=20,\n",
    "                              validation_data=val_gen,\n",
    "                              validation_steps=val_steps)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "\n",
    "loss = history.history['loss']\n",
    "val_loss = history.history['val_loss']\n",
    "\n",
    "epochs = range(1, len(loss) + 1)\n",
    "\n",
    "plt.figure()\n",
    "plt.plot(epochs, loss, 'bo', label='Training loss')\n",
    "plt.plot(epochs, val_loss, 'b', label='Validation loss')\n",
    "plt.title('Basic model loss')\n",
    "plt.legend()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 1  GRU"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/20\n",
      "500/500 [==============================] - 164s 328ms/step - loss: 0.3020 - val_loss: 0.2684\n",
      "Epoch 2/20\n",
      "500/500 [==============================] - 175s 351ms/step - loss: 0.2862 - val_loss: 0.2672\n",
      "Epoch 3/20\n",
      "500/500 [==============================] - 166s 333ms/step - loss: 0.2812 - val_loss: 0.2653\n",
      "Epoch 4/20\n",
      "500/500 [==============================] - 165s 330ms/step - loss: 0.2752 - val_loss: 0.2643\n",
      "Epoch 5/20\n",
      "500/500 [==============================] - 161s 322ms/step - loss: 0.2706 - val_loss: 0.2700\n",
      "Epoch 6/20\n",
      "500/500 [==============================] - 174s 347ms/step - loss: 0.2661 - val_loss: 0.2693\n",
      "Epoch 7/20\n",
      "500/500 [==============================] - 167s 333ms/step - loss: 0.2619 - val_loss: 0.2708\n",
      "Epoch 8/20\n",
      "500/500 [==============================] - 160s 320ms/step - loss: 0.2556 - val_loss: 0.2732\n",
      "Epoch 9/20\n",
      "500/500 [==============================] - 162s 323ms/step - loss: 0.2503 - val_loss: 0.2745\n",
      "Epoch 10/20\n",
      "500/500 [==============================] - 159s 319ms/step - loss: 0.2461 - val_loss: 0.2780\n",
      "Epoch 11/20\n",
      "500/500 [==============================] - 163s 326ms/step - loss: 0.2404 - val_loss: 0.2809\n",
      "Epoch 12/20\n",
      "500/500 [==============================] - 164s 327ms/step - loss: 0.2369 - val_loss: 0.2866\n",
      "Epoch 13/20\n",
      "500/500 [==============================] - 158s 316ms/step - loss: 0.2321 - val_loss: 0.2862\n",
      "Epoch 14/20\n",
      "500/500 [==============================] - 1110s 2s/step - loss: 0.2285 - val_loss: 0.2871\n",
      "Epoch 15/20\n",
      "500/500 [==============================] - 133s 266ms/step - loss: 0.2257 - val_loss: 0.2889\n",
      "Epoch 16/20\n",
      "500/500 [==============================] - 131s 263ms/step - loss: 0.2206 - val_loss: 0.2908\n",
      "Epoch 17/20\n",
      "500/500 [==============================] - 133s 266ms/step - loss: 0.2181 - val_loss: 0.2965\n",
      "Epoch 18/20\n",
      "500/500 [==============================] - 150s 301ms/step - loss: 0.2147 - val_loss: 0.2958\n",
      "Epoch 19/20\n",
      "500/500 [==============================] - 153s 306ms/step - loss: 0.2110 - val_loss: 0.2995\n",
      "Epoch 20/20\n",
      "500/500 [==============================] - 137s 274ms/step - loss: 0.2082 - val_loss: 0.2995\n"
     ]
    }
   ],
   "source": [
    "model = Sequential()\n",
    "model.add(layers.GRU(32, input_shape = (None, float_data.shape[-1])))\n",
    "model.add(layers.Dense(1))\n",
    "\n",
    "model.compile(optimizer=RMSprop(), loss='mae')\n",
    "history = model.fit_generator(train_gen,\n",
    "                              steps_per_epoch=500,\n",
    "                              epochs=20,\n",
    "                              validation_data=val_gen,\n",
    "                              validation_steps=val_steps)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "loss = history.history['loss']\n",
    "val_loss = history.history['val_loss']\n",
    "\n",
    "epochs = range(1, len(loss) + 1)\n",
    "\n",
    "plt.figure()\n",
    "plt.plot(epochs, loss, 'bo', label='Training loss')\n",
    "plt.plot(epochs, val_loss, 'b', label='Validation loss')\n",
    "plt.title('GRU model loss')\n",
    "plt.legend()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Use Dropout"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/40\n",
      "500/500 [==============================] - 204s 408ms/step - loss: 0.3455 - val_loss: 0.2750\n",
      "Epoch 2/40\n",
      "500/500 [==============================] - 191s 383ms/step - loss: 0.3143 - val_loss: 0.2726\n",
      "Epoch 3/40\n",
      "500/500 [==============================] - 194s 388ms/step - loss: 0.3077 - val_loss: 0.2698\n",
      "Epoch 4/40\n",
      "500/500 [==============================] - 190s 380ms/step - loss: 0.3051 - val_loss: 0.2701\n",
      "Epoch 5/40\n",
      "500/500 [==============================] - 193s 385ms/step - loss: 0.3007 - val_loss: 0.2703\n",
      "Epoch 6/40\n",
      "500/500 [==============================] - 193s 386ms/step - loss: 0.2970 - val_loss: 0.2681\n",
      "Epoch 7/40\n",
      "500/500 [==============================] - 192s 384ms/step - loss: 0.2947 - val_loss: 0.2704\n",
      "Epoch 8/40\n",
      "500/500 [==============================] - 202s 404ms/step - loss: 0.2936 - val_loss: 0.2662\n",
      "Epoch 9/40\n",
      "500/500 [==============================] - 177s 354ms/step - loss: 0.2931 - val_loss: 0.2675\n",
      "Epoch 10/40\n",
      "500/500 [==============================] - 182s 364ms/step - loss: 0.2908 - val_loss: 0.2649\n",
      "Epoch 11/40\n",
      "500/500 [==============================] - 194s 387ms/step - loss: 0.2890 - val_loss: 0.2658\n",
      "Epoch 12/40\n",
      "500/500 [==============================] - 190s 379ms/step - loss: 0.2879 - val_loss: 0.2669\n",
      "Epoch 13/40\n",
      "500/500 [==============================] - 187s 373ms/step - loss: 0.2864 - val_loss: 0.2656\n",
      "Epoch 14/40\n",
      "500/500 [==============================] - 187s 374ms/step - loss: 0.2861 - val_loss: 0.2660\n",
      "Epoch 15/40\n",
      "500/500 [==============================] - 182s 364ms/step - loss: 0.2867 - val_loss: 0.2658\n",
      "Epoch 16/40\n",
      "500/500 [==============================] - 183s 366ms/step - loss: 0.2831 - val_loss: 0.2615\n",
      "Epoch 17/40\n",
      "500/500 [==============================] - 177s 355ms/step - loss: 0.2841 - val_loss: 0.2612\n",
      "Epoch 18/40\n",
      "500/500 [==============================] - 190s 380ms/step - loss: 0.2812 - val_loss: 0.2650\n",
      "Epoch 19/40\n",
      "500/500 [==============================] - 214s 429ms/step - loss: 0.2811 - val_loss: 0.2639\n",
      "Epoch 20/40\n",
      "500/500 [==============================] - 185s 370ms/step - loss: 0.2793 - val_loss: 0.2630\n",
      "Epoch 21/40\n",
      "500/500 [==============================] - 221s 443ms/step - loss: 0.2793 - val_loss: 0.2626\n",
      "Epoch 22/40\n",
      "500/500 [==============================] - 190s 380ms/step - loss: 0.2792 - val_loss: 0.2628\n",
      "Epoch 23/40\n",
      "500/500 [==============================] - 202s 404ms/step - loss: 0.2771 - val_loss: 0.2637\n",
      "Epoch 24/40\n",
      "500/500 [==============================] - 198s 395ms/step - loss: 0.2780 - val_loss: 0.2636\n",
      "Epoch 25/40\n",
      "500/500 [==============================] - 190s 380ms/step - loss: 0.2769 - val_loss: 0.2615\n",
      "Epoch 26/40\n",
      "500/500 [==============================] - 194s 389ms/step - loss: 0.2756 - val_loss: 0.2645\n",
      "Epoch 27/40\n",
      "500/500 [==============================] - 201s 402ms/step - loss: 0.2746 - val_loss: 0.2614\n",
      "Epoch 28/40\n",
      "500/500 [==============================] - 189s 378ms/step - loss: 0.2748 - val_loss: 0.2629\n",
      "Epoch 29/40\n",
      "500/500 [==============================] - 195s 389ms/step - loss: 0.2738 - val_loss: 0.2665\n",
      "Epoch 30/40\n",
      "500/500 [==============================] - 200s 400ms/step - loss: 0.2720 - val_loss: 0.2631\n",
      "Epoch 31/40\n",
      "500/500 [==============================] - 194s 389ms/step - loss: 0.2737 - val_loss: 0.2637\n",
      "Epoch 32/40\n",
      "500/500 [==============================] - 201s 402ms/step - loss: 0.2721 - val_loss: 0.2627\n",
      "Epoch 33/40\n",
      "500/500 [==============================] - 193s 386ms/step - loss: 0.2737 - val_loss: 0.2643\n",
      "Epoch 34/40\n",
      "500/500 [==============================] - 186s 372ms/step - loss: 0.2724 - val_loss: 0.2641\n",
      "Epoch 35/40\n",
      "500/500 [==============================] - 185s 369ms/step - loss: 0.2708 - val_loss: 0.2660\n",
      "Epoch 36/40\n",
      "500/500 [==============================] - 194s 388ms/step - loss: 0.2707 - val_loss: 0.2633\n",
      "Epoch 37/40\n",
      "500/500 [==============================] - 187s 375ms/step - loss: 0.2707 - val_loss: 0.2641\n",
      "Epoch 38/40\n",
      "500/500 [==============================] - 191s 382ms/step - loss: 0.2697 - val_loss: 0.2614\n",
      "Epoch 39/40\n",
      "500/500 [==============================] - 190s 380ms/step - loss: 0.2700 - val_loss: 0.2649\n",
      "Epoch 40/40\n",
      "500/500 [==============================] - 181s 361ms/step - loss: 0.2679 - val_loss: 0.2706\n"
     ]
    }
   ],
   "source": [
    "model = Sequential()\n",
    "model.add(layers.GRU(32, \n",
    "                     dropout=0.2,\n",
    "                     recurrent_dropout=0.2,\n",
    "                     input_shape=(None, float_data.shape[-1])))\n",
    "model.add(layers.Dense(1))\n",
    "\n",
    "model.compile(optimizer=RMSprop(), loss='mae')\n",
    "history = model.fit_generator(train_gen,\n",
    "                              steps_per_epoch=500,\n",
    "                              epochs=40,\n",
    "                              validation_data=val_gen,\n",
    "                              validation_steps=val_steps)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "loss = history.history['loss']\n",
    "val_loss = history.history['val_loss']\n",
    "\n",
    "epochs = range(1, len(loss) + 1)\n",
    "\n",
    "plt.figure()\n",
    "plt.plot(epochs, loss, 'bo', label='Training loss')\n",
    "plt.plot(epochs, val_loss, 'b', label='Validation loss')\n",
    "plt.title('GRU + Dropout model loss')\n",
    "plt.legend()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## GRU + Dropout + Add layer model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/40\n",
      "500/500 [==============================] - 432s 864ms/step - loss: 0.3369 - val_loss: 0.2785\n",
      "Epoch 2/40\n",
      "500/500 [==============================] - 393s 785ms/step - loss: 0.3135 - val_loss: 0.2749\n",
      "Epoch 3/40\n",
      "500/500 [==============================] - 396s 792ms/step - loss: 0.3059 - val_loss: 0.2710\n",
      "Epoch 4/40\n",
      "500/500 [==============================] - 397s 795ms/step - loss: 0.3023 - val_loss: 0.2712\n",
      "Epoch 5/40\n",
      "500/500 [==============================] - 397s 793ms/step - loss: 0.2984 - val_loss: 0.2667\n",
      "Epoch 6/40\n",
      "500/500 [==============================] - 399s 797ms/step - loss: 0.2960 - val_loss: 0.2698\n",
      "Epoch 7/40\n",
      "500/500 [==============================] - 1059s 2s/step - loss: 0.2913 - val_loss: 0.2663\n",
      "Epoch 8/40\n",
      "500/500 [==============================] - 315s 630ms/step - loss: 0.2891 - val_loss: 0.2653\n",
      "Epoch 9/40\n",
      "500/500 [==============================] - 310s 620ms/step - loss: 0.2898 - val_loss: 0.2666\n",
      "Epoch 10/40\n",
      "500/500 [==============================] - 308s 616ms/step - loss: 0.2861 - val_loss: 0.2655\n",
      "Epoch 11/40\n",
      "500/500 [==============================] - 313s 626ms/step - loss: 0.2862 - val_loss: 0.2668\n",
      "Epoch 12/40\n",
      "500/500 [==============================] - 676s 1s/step - loss: 0.2844 - val_loss: 0.2666\n",
      "Epoch 13/40\n",
      "500/500 [==============================] - 313s 625ms/step - loss: 0.2817 - val_loss: 0.2650\n",
      "Epoch 14/40\n",
      "500/500 [==============================] - 313s 626ms/step - loss: 0.2824 - val_loss: 0.2688\n",
      "Epoch 15/40\n",
      "500/500 [==============================] - 313s 625ms/step - loss: 0.2808 - val_loss: 0.2664\n",
      "Epoch 16/40\n",
      "500/500 [==============================] - 314s 627ms/step - loss: 0.2765 - val_loss: 0.2707\n",
      "Epoch 17/40\n",
      "500/500 [==============================] - 315s 630ms/step - loss: 0.2761 - val_loss: 0.2674\n",
      "Epoch 18/40\n",
      "500/500 [==============================] - 316s 631ms/step - loss: 0.2752 - val_loss: 0.2695\n",
      "Epoch 19/40\n",
      "500/500 [==============================] - 316s 633ms/step - loss: 0.2714 - val_loss: 0.2705\n",
      "Epoch 20/40\n",
      "500/500 [==============================] - 316s 632ms/step - loss: 0.2712 - val_loss: 0.2652\n",
      "Epoch 21/40\n",
      "500/500 [==============================] - 311s 623ms/step - loss: 0.2711 - val_loss: 0.2664\n",
      "Epoch 22/40\n",
      "500/500 [==============================] - 308s 617ms/step - loss: 0.2676 - val_loss: 0.2753\n",
      "Epoch 23/40\n",
      "500/500 [==============================] - 309s 618ms/step - loss: 0.2676 - val_loss: 0.2737\n",
      "Epoch 24/40\n",
      "500/500 [==============================] - 308s 616ms/step - loss: 0.2658 - val_loss: 0.2658\n",
      "Epoch 25/40\n",
      "500/500 [==============================] - 307s 615ms/step - loss: 0.2639 - val_loss: 0.2660\n",
      "Epoch 26/40\n",
      "500/500 [==============================] - 308s 616ms/step - loss: 0.2635 - val_loss: 0.2677\n",
      "Epoch 27/40\n",
      "500/500 [==============================] - 310s 620ms/step - loss: 0.2627 - val_loss: 0.2709\n",
      "Epoch 28/40\n",
      "500/500 [==============================] - 308s 616ms/step - loss: 0.2610 - val_loss: 0.2649\n",
      "Epoch 29/40\n",
      "500/500 [==============================] - 308s 616ms/step - loss: 0.2611 - val_loss: 0.2663\n",
      "Epoch 30/40\n",
      "500/500 [==============================] - 309s 617ms/step - loss: 0.2586 - val_loss: 0.2795\n",
      "Epoch 31/40\n",
      "500/500 [==============================] - 309s 618ms/step - loss: 0.2593 - val_loss: 0.2755\n",
      "Epoch 32/40\n",
      "500/500 [==============================] - 308s 617ms/step - loss: 0.2591 - val_loss: 0.2690\n",
      "Epoch 33/40\n",
      "500/500 [==============================] - 309s 617ms/step - loss: 0.2568 - val_loss: 0.2708\n",
      "Epoch 34/40\n",
      "500/500 [==============================] - 310s 619ms/step - loss: 0.2574 - val_loss: 0.2711\n",
      "Epoch 35/40\n",
      "500/500 [==============================] - 314s 628ms/step - loss: 0.2560 - val_loss: 0.2775\n",
      "Epoch 36/40\n",
      "500/500 [==============================] - 313s 627ms/step - loss: 0.2567 - val_loss: 0.2743\n",
      "Epoch 37/40\n",
      "500/500 [==============================] - 313s 626ms/step - loss: 0.2551 - val_loss: 0.2760\n",
      "Epoch 38/40\n",
      "500/500 [==============================] - 314s 628ms/step - loss: 0.2555 - val_loss: 0.2781\n",
      "Epoch 39/40\n",
      "500/500 [==============================] - 323s 646ms/step - loss: 0.2556 - val_loss: 0.2790\n",
      "Epoch 40/40\n",
      "500/500 [==============================] - 315s 630ms/step - loss: 0.2534 - val_loss: 0.2717\n"
     ]
    }
   ],
   "source": [
    "model = Sequential()\n",
    "model.add(layers.GRU(32, \n",
    "                     dropout=0.1,\n",
    "                     recurrent_dropout=0.5,\n",
    "                     return_sequences=True,\n",
    "                     input_shape=(None, float_data.shape[-1])))\n",
    "model.add(layers.GRU(64, activation='relu',\n",
    "                     dropout=0.1,\n",
    "                     recurrent_dropout=0.5))\n",
    "model.add(layers.Dense(1))\n",
    "\n",
    "model.compile(optimizer=RMSprop(), loss='mae')\n",
    "history = model.fit_generator(train_gen,\n",
    "                              steps_per_epoch=500,\n",
    "                              epochs=40,\n",
    "                              validation_data=val_gen,\n",
    "                              validation_steps=val_steps)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "loss = history.history['loss']\n",
    "val_loss = history.history['val_loss']\n",
    "\n",
    "epochs = range(1, len(loss) + 1)\n",
    "\n",
    "plt.figure()\n",
    "plt.plot(epochs, loss, 'bo', label='Training loss')\n",
    "plt.plot(epochs, val_loss, 'b', label='Validation loss')\n",
    "plt.title('GRU + Dropout + Add layer model loss')\n",
    "plt.legend()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## bidirectional GRU"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/40\n",
      "500/500 [==============================] - 268s 536ms/step - loss: 0.2962 - val_loss: 0.2985\n",
      "Epoch 2/40\n",
      "500/500 [==============================] - 267s 534ms/step - loss: 0.2747 - val_loss: 0.2701\n",
      "Epoch 3/40\n",
      "500/500 [==============================] - 267s 534ms/step - loss: 0.2661 - val_loss: 0.2661\n",
      "Epoch 4/40\n",
      "500/500 [==============================] - 266s 531ms/step - loss: 0.2603 - val_loss: 0.2688\n",
      "Epoch 5/40\n",
      "500/500 [==============================] - 265s 531ms/step - loss: 0.2551 - val_loss: 0.2726\n",
      "Epoch 6/40\n",
      "500/500 [==============================] - 266s 531ms/step - loss: 0.2507 - val_loss: 0.2686\n",
      "Epoch 7/40\n",
      "500/500 [==============================] - 262s 524ms/step - loss: 0.2456 - val_loss: 0.2721\n",
      "Epoch 8/40\n",
      "500/500 [==============================] - 261s 521ms/step - loss: 0.2393 - val_loss: 0.2747\n",
      "Epoch 9/40\n",
      "500/500 [==============================] - 260s 521ms/step - loss: 0.2337 - val_loss: 0.2802\n",
      "Epoch 10/40\n",
      "500/500 [==============================] - 261s 521ms/step - loss: 0.2277 - val_loss: 0.2860\n",
      "Epoch 11/40\n",
      "500/500 [==============================] - 260s 521ms/step - loss: 0.2227 - val_loss: 0.2884\n",
      "Epoch 12/40\n",
      "500/500 [==============================] - 260s 520ms/step - loss: 0.2158 - val_loss: 0.2936\n",
      "Epoch 13/40\n",
      "500/500 [==============================] - 260s 521ms/step - loss: 0.2102 - val_loss: 0.3008\n",
      "Epoch 14/40\n",
      "500/500 [==============================] - 261s 521ms/step - loss: 0.2041 - val_loss: 0.3037\n",
      "Epoch 15/40\n",
      "500/500 [==============================] - 261s 522ms/step - loss: 0.1996 - val_loss: 0.3036\n",
      "Epoch 16/40\n",
      "500/500 [==============================] - 260s 520ms/step - loss: 0.1948 - val_loss: 0.3069\n",
      "Epoch 17/40\n",
      "500/500 [==============================] - 260s 520ms/step - loss: 0.1887 - val_loss: 0.3103\n",
      "Epoch 18/40\n",
      "500/500 [==============================] - 260s 519ms/step - loss: 0.1865 - val_loss: 0.3059\n",
      "Epoch 19/40\n",
      "500/500 [==============================] - 260s 521ms/step - loss: 0.1812 - val_loss: 0.3116\n",
      "Epoch 20/40\n",
      "500/500 [==============================] - 260s 521ms/step - loss: 0.1799 - val_loss: 0.3191\n",
      "Epoch 21/40\n",
      "500/500 [==============================] - 261s 522ms/step - loss: 0.1750 - val_loss: 0.3224\n",
      "Epoch 22/40\n",
      "500/500 [==============================] - 265s 530ms/step - loss: 0.1716 - val_loss: 0.3194\n",
      "Epoch 23/40\n",
      "500/500 [==============================] - 264s 529ms/step - loss: 0.1681 - val_loss: 0.3190\n",
      "Epoch 24/40\n",
      "500/500 [==============================] - 263s 527ms/step - loss: 0.1662 - val_loss: 0.3260\n",
      "Epoch 25/40\n",
      "500/500 [==============================] - 264s 528ms/step - loss: 0.1636 - val_loss: 0.3309\n",
      "Epoch 26/40\n",
      "500/500 [==============================] - 266s 531ms/step - loss: 0.1598 - val_loss: 0.3310\n",
      "Epoch 27/40\n",
      "500/500 [==============================] - 268s 536ms/step - loss: 0.1586 - val_loss: 0.3237\n",
      "Epoch 28/40\n",
      "500/500 [==============================] - 267s 534ms/step - loss: 0.1559 - val_loss: 0.3258\n",
      "Epoch 29/40\n",
      "500/500 [==============================] - 267s 534ms/step - loss: 0.1537 - val_loss: 0.3290\n",
      "Epoch 30/40\n",
      "500/500 [==============================] - 266s 532ms/step - loss: 0.1519 - val_loss: 0.3326\n",
      "Epoch 31/40\n",
      "500/500 [==============================] - 265s 530ms/step - loss: 0.1501 - val_loss: 0.3322\n",
      "Epoch 32/40\n",
      "500/500 [==============================] - 265s 530ms/step - loss: 0.1479 - val_loss: 0.3293\n",
      "Epoch 33/40\n",
      "500/500 [==============================] - 266s 532ms/step - loss: 0.1461 - val_loss: 0.3340\n",
      "Epoch 34/40\n",
      "500/500 [==============================] - 266s 533ms/step - loss: 0.1448 - val_loss: 0.3344\n",
      "Epoch 35/40\n",
      "500/500 [==============================] - 264s 529ms/step - loss: 0.1434 - val_loss: 0.3302\n",
      "Epoch 36/40\n",
      "500/500 [==============================] - 261s 522ms/step - loss: 0.1419 - val_loss: 0.3329\n",
      "Epoch 37/40\n",
      "500/500 [==============================] - 260s 520ms/step - loss: 0.1411 - val_loss: 0.3384\n",
      "Epoch 38/40\n",
      "500/500 [==============================] - 260s 519ms/step - loss: 0.1391 - val_loss: 0.3336\n",
      "Epoch 39/40\n",
      "500/500 [==============================] - 261s 521ms/step - loss: 0.1373 - val_loss: 0.3390\n",
      "Epoch 40/40\n",
      "500/500 [==============================] - 260s 521ms/step - loss: 0.1362 - val_loss: 0.3335\n"
     ]
    }
   ],
   "source": [
    "model = Sequential()\n",
    "model.add(layers.Bidirectional(\n",
    "    layers.GRU(32), input_shape=(None, float_data.shape[-1])))\n",
    "model.add(layers.Dense(1))\n",
    "\n",
    "model.compile(optimizer='RMSprop', loss='mae')\n",
    "history = model.fit_generator(train_gen,\n",
    "                              steps_per_epoch=500,\n",
    "                              epochs=40,\n",
    "                              validation_data=val_gen,\n",
    "                              validation_steps=val_steps)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "loss = history.history['loss']\n",
    "val_loss = history.history['val_loss']\n",
    "\n",
    "epochs = range(1, len(loss) + 1)\n",
    "\n",
    "plt.figure()\n",
    "plt.plot(epochs, loss, 'bo', label='Training loss')\n",
    "plt.plot(epochs, val_loss, 'b', label='Validation loss')\n",
    "plt.title('bidirectional GRU model loss')\n",
    "plt.legend()\n",
    "\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "tensorflowGPU",
   "language": "python",
   "name": "tensorflow"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.5.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}