imdb.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Imdb sentiment classification.\n",
    "\n",
    "Dataset of 25,000 movies reviews from IMDB, labeled by sentiment (positive/negative). Reviews have been preprocessed, and each review is encoded as a sequence of word indexes (integers). For convenience, words are indexed by overall frequency in the dataset, so that for instance the integer \"3\" encodes the 3rd most frequent word in the data. This allows for quick filtering operations such as: \"only consider the top 10,000 most common words, but eliminate the top 20 most common words\"."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Basic packages.\n",
    "import os\n",
    "import sys\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.metrics import classification_report\n",
    "\n",
    "# Keras specific packages.\n",
    "from keras import Input\n",
    "from keras import Model\n",
    "from keras import regularizers\n",
    "from keras import optimizers\n",
    "from keras.layers import Dense, Activation, Flatten, GRU\n",
    "from keras.layers import Dropout\n",
    "from keras.layers import Conv1D, MaxPooling1D\n",
    "from keras.layers import Embedding\n",
    "from keras.preprocessing.text import Tokenizer\n",
    "from keras.preprocessing.sequence import pad_sequences\n",
    "from keras.utils import to_categorical\n",
    "from keras.datasets import imdb"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "MAX_NUM_WORDS = 10000\n",
    "MAX_SEQUENCE_LENGTH = 1000\n",
    "EMBEDDING_DIM = 100\n",
    "VALIDATION_SPLIT = 0.25\n",
    "TEXT_DATA_DIR = \"dataset/20_newsgroup\"\n",
    "GLOVE_DIR = \"dataset/glove\"\n",
    "EPOCHS = 10\n",
    "BATCH_SIZE = 129"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 1. Load the dataset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load the data.\n",
    "(x_train, y_train), (x_test, y_test) = imdb.load_data(num_words=MAX_NUM_WORDS)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Get the word to index dict.\n",
    "word_to_index = imdb.get_word_index()\n",
    "\n",
    "# Get the index to word dict.\n",
    "index_to_word = dict(\n",
    "[(value, key) for (key, value) in word_to_index.items()])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Length dictionnary = 88584\n",
      "2494\n"
     ]
    }
   ],
   "source": [
    "# Display\n",
    "print(\"Length dictionnary = {}\".format(len(word_to_index)))\n",
    "\n",
    "max_row = []\n",
    "for i in range(x_train.shape[0]):\n",
    "    max_row.append(len(x_train[i]))\n",
    "print(max(max_row))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2. Preparing the pretrained embedding layer."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 400000 word vectors.\n"
     ]
    }
   ],
   "source": [
    "embeddings_index = {}\n",
    "f = open(os.path.join(GLOVE_DIR, \"glove.6B.{}d.txt\".format(EMBEDDING_DIM)))\n",
    "for line in f:\n",
    "    values = line.split()\n",
    "    word = values[0]\n",
    "    coefs = np.asarray(values[1:], dtype=\"float32\")\n",
    "    embeddings_index[word] = coefs\n",
    "f.close()\n",
    "\n",
    "print(\"Found %s word vectors.\" % len(embeddings_index))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [],
   "source": [
    "embedding_matrix = np.zeros((len(word_to_index) + 1, EMBEDDING_DIM))\n",
    "for word, i in word_to_index.items():\n",
    "    embedding_vector = embeddings_index.get(word)\n",
    "    if embedding_vector is not None:\n",
    "        # words not found in embedding index will be all-zeros.\n",
    "        embedding_matrix[i] = embedding_vector"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [],
   "source": [
    "embedding_layer = Embedding(len(word_to_index) + 1,\n",
    "                            EMBEDDING_DIM,\n",
    "                            weights=[embedding_matrix],\n",
    "                            input_length=MAX_SEQUENCE_LENGTH,\n",
    "                            trainable=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3. Handle the dataset.\n",
    "\n",
    "Here we gather the features of words $X \\in \\mathbb{R}^{m \\times n}$ where $m$ is the total number of samples and $n$ is the features length. For the current example $n$ is equal to $10000$.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Size x_tr = (25000, 1000)\n",
      "Size x_te = (25000, 1000)\n"
     ]
    }
   ],
   "source": [
    "# Pad the training and test features.\n",
    "x_tr = pad_sequences(x_train, maxlen=MAX_SEQUENCE_LENGTH)\n",
    "x_te = pad_sequences(x_test, maxlen=MAX_SEQUENCE_LENGTH)\n",
    "\n",
    "# Display the size.\n",
    "print(\"Size x_tr = {}\".format(x_tr.shape))\n",
    "print(\"Size x_te = {}\".format(x_te.shape))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "y_train  (25000, 1)\n",
      "y_test  (25000, 1)\n"
     ]
    }
   ],
   "source": [
    "# Handle the training and test labels.\n",
    "y_tr = y_train.reshape(-1, 1)\n",
    "y_te = y_test.reshape(-1, 1)\n",
    "\n",
    "# Display the shapes.\n",
    "print(\"y_train \", y_tr.shape)\n",
    "print(\"y_test \", y_te.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3. Build the model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_14 (InputLayer)        (None, 1000)              0         \n",
      "_________________________________________________________________\n",
      "embedding_3 (Embedding)      (None, 1000, 100)         8858500   \n",
      "_________________________________________________________________\n",
      "gru_2 (GRU)                  (None, 128)               87936     \n",
      "_________________________________________________________________\n",
      "dense_15 (Dense)             (None, 1)                 129       \n",
      "=================================================================\n",
      "Total params: 8,946,565\n",
      "Trainable params: 88,065\n",
      "Non-trainable params: 8,858,500\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# Set the input.\n",
    "sequence_input = Input(shape=(MAX_SEQUENCE_LENGTH,), dtype=\"int32\")\n",
    "\n",
    "# Set the embedding layer.\n",
    "embedded_sequences = embedding_layer(sequence_input)\n",
    "\n",
    "# Conv layer 1.\n",
    "\"\"\"\n",
    "x = Conv1D(64, 5, kernel_regularizer=regularizers.l2(0.001))(embedded_sequences)\n",
    "x = Activation(\"relu\")(x)\n",
    "x = MaxPooling1D(5)(x)\n",
    "X = Dropout(0.5)(x)\n",
    "\n",
    "# Conv Layer 2.\n",
    "x = Conv1D(64, 5, kernel_regularizer=regularizers.l2(0.001))(x)\n",
    "x = Activation(\"relu\")(x)\n",
    "x = MaxPooling1D(5)(x)\n",
    "X = Dropout(0.5)(x)\n",
    "\n",
    "# Conv Layer 3.\n",
    "x = Conv1D(64, 5, kernel_regularizer=regularizers.l2(0.001))(x)\n",
    "x = Activation(\"relu\")(x)\n",
    "x = MaxPooling1D(35)(x)\n",
    "X = Dropout(0.5)(x)\n",
    "\n",
    "# Output layer.\n",
    "x = Flatten()(x)\n",
    "x = Dense(128)(x)\n",
    "x = Activation(\"relu\")(x)\n",
    "X = Dropout(0.5)(x)\n",
    "\n",
    "\"\"\"\n",
    "\n",
    "#x = Flatten()(x)\n",
    "#x = Dense(128)(x)\n",
    "#x = Activation(\"relu\")(x)\n",
    "#X = Dropout(0.5)(x)\n",
    "\n",
    "x = GRU(128, return_sequences=False)(embedded_sequences)\n",
    "\n",
    "# Softmax layer.\n",
    "preds = Dense(1, activation=\"sigmoid\")(x)\n",
    "\n",
    "# Build the model.\n",
    "model = Model(sequence_input, preds)\n",
    "\n",
    "# Set the optimizer.\n",
    "optim = optimizers.Adam(lr=0.001)\n",
    "\n",
    "# Compile the model.\n",
    "model.compile(loss=\"binary_crossentropy\", optimizer=optim, metrics=[\"acc\"])\n",
    "\n",
    "# Set the fitting parameters.\n",
    "fit_params = {\n",
    "    \"epochs\": EPOCHS,\n",
    "    \"batch_size\": BATCH_SIZE,\n",
    "    \"validation_split\": VALIDATION_SPLIT,\n",
    "    \"shuffle\": True\n",
    "}\n",
    "\n",
    "# Print the model.\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 18750 samples, validate on 6250 samples\n",
      "Epoch 1/10\n",
      "18750/18750 [==============================] - 300s 16ms/step - loss: 0.6888 - acc: 0.5370 - val_loss: 0.6762 - val_acc: 0.5653\n",
      "Epoch 2/10\n",
      " 1677/18750 [=>............................] - ETA: 3:44 - loss: 0.6670 - acc: 0.5951"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-66-642371bf3afc>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0;31m# Fit the model.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 2\u001b[0;31m \u001b[0mhistory\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mmodel\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx_tr\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0my_tr\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mfit_params\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[0;32m~/miniconda3/lib/python3.6/site-packages/keras/engine/training.py\u001b[0m in \u001b[0;36mfit\u001b[0;34m(self, x, y, batch_size, epochs, verbose, callbacks, validation_split, validation_data, shuffle, class_weight, sample_weight, initial_epoch, steps_per_epoch, validation_steps, **kwargs)\u001b[0m\n\u001b[1;32m   1703\u001b[0m                               \u001b[0minitial_epoch\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0minitial_epoch\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1704\u001b[0m                               \u001b[0msteps_per_epoch\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0msteps_per_epoch\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1705\u001b[0;31m                               validation_steps=validation_steps)\n\u001b[0m\u001b[1;32m   1706\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1707\u001b[0m     def evaluate(self, x=None, y=None,\n",
      "\u001b[0;32m~/miniconda3/lib/python3.6/site-packages/keras/engine/training.py\u001b[0m in \u001b[0;36m_fit_loop\u001b[0;34m(self, f, ins, out_labels, batch_size, epochs, verbose, callbacks, val_f, val_ins, shuffle, callback_metrics, initial_epoch, steps_per_epoch, validation_steps)\u001b[0m\n\u001b[1;32m   1233\u001b[0m                         \u001b[0mins_batch\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mi\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mins_batch\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mi\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtoarray\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1234\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1235\u001b[0;31m                     \u001b[0mouts\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mf\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mins_batch\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1236\u001b[0m                     \u001b[0;32mif\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0misinstance\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mouts\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlist\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1237\u001b[0m                         \u001b[0mouts\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0mouts\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/miniconda3/lib/python3.6/site-packages/keras/backend/tensorflow_backend.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, inputs)\u001b[0m\n\u001b[1;32m   2476\u001b[0m         \u001b[0msession\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mget_session\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2477\u001b[0m         updated = session.run(fetches=fetches, feed_dict=feed_dict,\n\u001b[0;32m-> 2478\u001b[0;31m                               **self.session_kwargs)\n\u001b[0m\u001b[1;32m   2479\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mupdated\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0mlen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0moutputs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2480\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/miniconda3/lib/python3.6/site-packages/tensorflow/python/client/session.py\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self, fetches, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m    898\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    899\u001b[0m       result = self._run(None, fetches, feed_dict, options_ptr,\n\u001b[0;32m--> 900\u001b[0;31m                          run_metadata_ptr)\n\u001b[0m\u001b[1;32m    901\u001b[0m       \u001b[0;32mif\u001b[0m \u001b[0mrun_metadata\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    902\u001b[0m         \u001b[0mproto_data\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtf_session\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mTF_GetBuffer\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mrun_metadata_ptr\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/miniconda3/lib/python3.6/site-packages/tensorflow/python/client/session.py\u001b[0m in \u001b[0;36m_run\u001b[0;34m(self, handle, fetches, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m   1133\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mfinal_fetches\u001b[0m \u001b[0;32mor\u001b[0m \u001b[0mfinal_targets\u001b[0m \u001b[0;32mor\u001b[0m \u001b[0;34m(\u001b[0m\u001b[0mhandle\u001b[0m \u001b[0;32mand\u001b[0m \u001b[0mfeed_dict_tensor\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1134\u001b[0m       results = self._do_run(handle, final_targets, final_fetches,\n\u001b[0;32m-> 1135\u001b[0;31m                              feed_dict_tensor, options, run_metadata)\n\u001b[0m\u001b[1;32m   1136\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1137\u001b[0m       \u001b[0mresults\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/miniconda3/lib/python3.6/site-packages/tensorflow/python/client/session.py\u001b[0m in \u001b[0;36m_do_run\u001b[0;34m(self, handle, target_list, fetch_list, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m   1314\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mhandle\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1315\u001b[0m       return self._do_call(_run_fn, feeds, fetches, targets, options,\n\u001b[0;32m-> 1316\u001b[0;31m                            run_metadata)\n\u001b[0m\u001b[1;32m   1317\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1318\u001b[0m       \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_do_call\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0m_prun_fn\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mhandle\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfeeds\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetches\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/miniconda3/lib/python3.6/site-packages/tensorflow/python/client/session.py\u001b[0m in \u001b[0;36m_do_call\u001b[0;34m(self, fn, *args)\u001b[0m\n\u001b[1;32m   1320\u001b[0m   \u001b[0;32mdef\u001b[0m \u001b[0m_do_call\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfn\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1321\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1322\u001b[0;31m       \u001b[0;32mreturn\u001b[0m \u001b[0mfn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1323\u001b[0m     \u001b[0;32mexcept\u001b[0m \u001b[0merrors\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mOpError\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1324\u001b[0m       \u001b[0mmessage\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcompat\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mas_text\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmessage\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/miniconda3/lib/python3.6/site-packages/tensorflow/python/client/session.py\u001b[0m in \u001b[0;36m_run_fn\u001b[0;34m(feed_dict, fetch_list, target_list, options, run_metadata)\u001b[0m\n\u001b[1;32m   1305\u001b[0m       \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_extend_graph\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1306\u001b[0m       return self._call_tf_sessionrun(\n\u001b[0;32m-> 1307\u001b[0;31m           options, feed_dict, fetch_list, target_list, run_metadata)\n\u001b[0m\u001b[1;32m   1308\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1309\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m_prun_fn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mhandle\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfeed_dict\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetch_list\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/miniconda3/lib/python3.6/site-packages/tensorflow/python/client/session.py\u001b[0m in \u001b[0;36m_call_tf_sessionrun\u001b[0;34m(self, options, feed_dict, fetch_list, target_list, run_metadata)\u001b[0m\n\u001b[1;32m   1407\u001b[0m       return tf_session.TF_SessionRun_wrapper(\n\u001b[1;32m   1408\u001b[0m           \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_session\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0moptions\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfeed_dict\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetch_list\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtarget_list\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1409\u001b[0;31m           run_metadata)\n\u001b[0m\u001b[1;32m   1410\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1411\u001b[0m       \u001b[0;32mwith\u001b[0m \u001b[0merrors\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mraise_exception_on_not_ok_status\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0mstatus\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "# Fit the model.\n",
    "history = model.fit(x_tr, y_tr, **fit_params)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x360 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Visualise the training resuls.\n",
    "plt.figure(figsize=(15,5))\n",
    "plt.subplot(121)\n",
    "plt.plot(history.history[\"loss\"], color=\"b\", label=\"tr\")\n",
    "plt.plot(history.history[\"val_loss\"], color=\"r\", label=\"te\")\n",
    "plt.ylabel(\"loss\")\n",
    "plt.xlabel(\"epochs\")\n",
    "plt.grid()\n",
    "plt.legend()\n",
    "plt.subplot(122)\n",
    "plt.plot(history.history[\"acc\"], color=\"b\", label=\"tr\")\n",
    "plt.plot(history.history[\"val_acc\"], color=\"r\", label=\"te\")\n",
    "plt.ylabel(\"acc\")\n",
    "plt.xlabel(\"epochs\")\n",
    "plt.grid()\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4. Evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "             precision    recall  f1-score   support\n",
      "\n",
      "          0       0.79      0.69      0.74     12500\n",
      "          1       0.72      0.82      0.77     12500\n",
      "\n",
      "avg / total       0.76      0.75      0.75     25000\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Get the predictions for the test dataset.\n",
    "y_pred = model.predict(x_te)\n",
    "\n",
    "# Update the predictions.\n",
    "y_pred = 1.0 * (y_pred > 0.5 )\n",
    "\n",
    "# Display the classification report.\n",
    "print(classification_report(y_te, y_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}