{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from keras.preprocessing import sequence\n",
    "from keras.utils import np_utils\n",
    "from keras.models import Sequential\n",
    "from keras.layers.core import Dense, Dropout, Activation\n",
    "from keras.layers.embeddings import Embedding\n",
    "from keras.layers.recurrent import LSTM\n",
    "from keras.datasets import imdb"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# IMDB Movie reviews sentiment classification - Keras (Backend: TensorFlow)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## IMDB Movie reviews Data\n",
    "\n",
    "Kerasの提供するデータセットを利用：  \n",
    "http://keras.io/datasets/#imdb-movie-reviews-sentiment-classification"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Downloading data from https://s3.amazonaws.com/text-datasets/imdb.pkl\n",
      "33218560/33213513 [==============================] - 11s    \n",
      "train sequences: 20000\n",
      "test sequences: 5000\n"
     ]
    }
   ],
   "source": [
    "max_features = 20000\n",
    "(X_train, y_train), (X_test, y_test) = imdb.load_data(nb_words=max_features, test_split=0.2)\n",
    "\n",
    "print 'train sequences: {0}'.format(len(X_train))\n",
    "print 'test sequences: {0}'.format(len(X_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[1, 20, 28, 716, 48, 495, 79, 27, 493, 8]"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# データの中身\n",
    "# 各レビューはWord Indexのシーケンスとして符号化されている\n",
    "# Indexは頻度に基づき符号化されているため，例えば3番めに頻度の高いWordにはIndex「3」が付与される\n",
    "\n",
    "X_train[0][:10]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X_train shape: (20000, 100)\n",
      "X_test shape: (5000, 100)\n"
     ]
    }
   ],
   "source": [
    "# テキストは最長100におさめる\n",
    "maxlen = 100\n",
    "\n",
    "X_train = sequence.pad_sequences(X_train, maxlen=maxlen)\n",
    "X_test = sequence.pad_sequences(X_test, maxlen=maxlen)\n",
    "\n",
    "print 'X_train shape: {0}'.format(X_train.shape)\n",
    "print 'X_test shape: {0}'.format(X_test.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Implementing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "model = Sequential()\n",
    "model.add(Embedding(max_features, 128, input_length=maxlen))\n",
    "model.add(LSTM(128))\n",
    "model.add(Dropout(0.5))\n",
    "model.add(Dense(1))\n",
    "model.add(Activation('sigmoid'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/amacbee/Dropbox/github/keras-conversational/lib/python2.7/site-packages/theano/scan_module/scan_perform_ext.py:133: RuntimeWarning: numpy.ndarray size changed, may indicate binary incompatibility\n",
      "  from scan_perform.scan_perform import *\n"
     ]
    }
   ],
   "source": [
    "model.compile(loss='binary_crossentropy', optimizer='adam', class_mode=\"binary\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 20000 samples, validate on 5000 samples\n",
      "Epoch 1/3\n",
      "20000/20000 [==============================] - 301s - loss: 0.4845 - acc: 0.7713 - val_loss: 0.4116 - val_acc: 0.8266\n",
      "Epoch 2/3\n",
      "20000/20000 [==============================] - 299s - loss: 0.2720 - acc: 0.8931 - val_loss: 0.3630 - val_acc: 0.8420\n",
      "Epoch 3/3\n",
      "20000/20000 [==============================] - 315s - loss: 0.1756 - acc: 0.9372 - val_loss: 0.4106 - val_acc: 0.8370\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x10bdc7f90>"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "batch_size = 32\n",
    "model.fit(X_train, y_train, batch_size=32, nb_epoch=3, validation_data=(X_test, y_test), show_accuracy=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Evaluating"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "5000/5000 [==============================] - 12s    \n",
      "[0.41059308743695982, 0.83699999999999997]\n"
     ]
    }
   ],
   "source": [
    "print(model.evaluate(X_test, y_test, batch_size=32, show_accuracy=True))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}