model_NN.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Movie Sentiment Analysis: Simple Neural Network"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "\n",
    "movie_reviews = pd.read_csv(\"data/IMDB_Dataset.csv\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Preprocess the data\n",
    "* Remove HTML tags\n",
    "* Remove non-alphabets\n",
    "* Remove single characters\n",
    "* Replace multiple spaces with single space"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import re\n",
    "import numpy as np\n",
    "\n",
    "tag_re = re.compile(r'<[^>]+>')\n",
    "def remove_tags(text):\n",
    "    return tag_re.sub('', text)\n",
    "\n",
    "def preprocess(sentence):\n",
    "    pp_sentence = remove_tags(sentence)\n",
    "    pp_sentence = re.sub('[^a-zA-Z]', ' ', pp_sentence)\n",
    "    pp_sentence = re.sub(r\"\\s+[a-zA-Z]\\s+\", ' ', pp_sentence)\n",
    "    pp_sentence = re.sub(r'\\s+', ' ', pp_sentence)\n",
    "    return pp_sentence\n",
    "\n",
    "review_sentences = list(movie_reviews['review']) # list the reviews\n",
    "\n",
    "X = []\n",
    "for sentence in review_sentences:\n",
    "    X.append(preprocess(sentence))\n",
    "    \n",
    "review_labels = list(map(lambda s: 1 if s==\"positive\" else 0, movie_reviews['sentiment'])) # list the sentiments\n",
    "\n",
    "y = np.array(review_labels)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Prepare training and testing data\n",
    "* 80% training data\n",
    "* 20% test data\n",
    "* seed 42"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.20, random_state=42)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Prepare Embedding layer"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Preprocess for the embedding layer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "from keras.preprocessing.text import Tokenizer\n",
    "from keras.preprocessing.sequence import pad_sequences\n",
    "\n",
    "tokenizer = Tokenizer(num_words=5000)\n",
    "tokenizer.fit_on_texts(X_train)\n",
    "\n",
    "X_train = tokenizer.texts_to_sequences(X_train)\n",
    "X_test = tokenizer.texts_to_sequences(X_test)\n",
    "\n",
    "vocab_size = len(tokenizer.word_index) + 1 # +1 because of reserved 0 index\n",
    "maxlen = 100\n",
    "\n",
    "X_train = pad_sequences(X_train, padding='post', maxlen=maxlen)\n",
    "X_test = pad_sequences(X_test, padding='post', maxlen=maxlen)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Create feature matrix using pretrained word embedding model\n",
    "* GloVe: http://nlp.stanford.edu/data/glove.6B.zip"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(92547, 100)\n"
     ]
    }
   ],
   "source": [
    "from numpy import asarray, zeros\n",
    "\n",
    "glove_file = open('data/glove.6B.100d.txt', encoding=\"utf8\")\n",
    "\n",
    "embedding_dictionary = dict()\n",
    "for line in glove_file:\n",
    "    records = line.split()\n",
    "    word = records[0]\n",
    "    vector = asarray(records[1:], dtype='float32')\n",
    "    embedding_dictionary[word] = vector\n",
    "    \n",
    "glove_file.close()\n",
    "\n",
    "embedding_matrix = zeros((vocab_size, 100))\n",
    "for word, index in tokenizer.word_index.items():\n",
    "    vector = embedding_dictionary.get(word)\n",
    "    if vector is not None:\n",
    "        embedding_matrix[index] = vector\n",
    "        \n",
    "print(embedding_matrix.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Text Classification model: Simple Neural Network\n",
    "* **Sequential model**\n",
    "* One **embedding layer**\n",
    "    * Since we are not training our own embeddings and using the GloVe embedding, we set trainable to False and in the weights attribute we pass our own embedding matrix.\n",
    "* **Flatten layer**\n",
    "* **Dense layer**\n",
    "    * Activation = sigmoid"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Create and train model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_1\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "embedding_1 (Embedding)      (None, 100, 100)          9254700   \n",
      "_________________________________________________________________\n",
      "flatten_1 (Flatten)          (None, 10000)             0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 1)                 10001     \n",
      "=================================================================\n",
      "Total params: 9,264,701\n",
      "Trainable params: 10,001\n",
      "Non-trainable params: 9,254,700\n",
      "_________________________________________________________________\n",
      "None\n",
      "Train on 32000 samples, validate on 8000 samples\n",
      "Epoch 1/6\n",
      "32000/32000 [==============================] - 1s 24us/step - loss: 0.6011 - acc: 0.6720 - val_loss: 0.5332 - val_acc: 0.7368\n",
      "Epoch 2/6\n",
      "32000/32000 [==============================] - 1s 20us/step - loss: 0.4921 - acc: 0.7640 - val_loss: 0.5151 - val_acc: 0.7458\n",
      "Epoch 3/6\n",
      "32000/32000 [==============================] - 1s 24us/step - loss: 0.4627 - acc: 0.7838 - val_loss: 0.5218 - val_acc: 0.7440\n",
      "Epoch 4/6\n",
      "32000/32000 [==============================] - 1s 26us/step - loss: 0.4353 - acc: 0.7990 - val_loss: 0.5301 - val_acc: 0.7415\n",
      "Epoch 5/6\n",
      "32000/32000 [==============================] - 1s 24us/step - loss: 0.4203 - acc: 0.8091 - val_loss: 0.5313 - val_acc: 0.7396\n",
      "Epoch 6/6\n",
      "32000/32000 [==============================] - 1s 19us/step - loss: 0.4093 - acc: 0.8147 - val_loss: 0.5422 - val_acc: 0.7385\n"
     ]
    }
   ],
   "source": [
    "from keras.models import Sequential\n",
    "from keras.layers.embeddings import Embedding\n",
    "from keras.layers import Flatten, Dense\n",
    "\n",
    "model = Sequential()\n",
    "embedding_layer = Embedding(vocab_size, 100, weights=[embedding_matrix], input_length=maxlen , trainable=False)\n",
    "\n",
    "model.add(embedding_layer)\n",
    "model.add(Flatten())\n",
    "model.add(Dense(1, activation='sigmoid'))\n",
    "\n",
    "model.compile(optimizer='adam', loss='binary_crossentropy', metrics=['acc'])\n",
    "print(model.summary())\n",
    "\n",
    "reviews_model = model.fit(X_train, y_train, batch_size=128, epochs=6, verbose=1, validation_split=0.2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Evaluate the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "10000/10000 [==============================] - 0s 30us/step\n",
      "Test Score: 54.85781580924988 %\n",
      "Test Accuracy: 73.58999848365784 %\n"
     ]
    }
   ],
   "source": [
    "score = model.evaluate(X_test, y_test, verbose=1)\n",
    "\n",
    "print(\"Test Score:\", score[0]*100, '%')\n",
    "print(\"Test Accuracy:\", score[1]*100, '%')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Plot Accuracy and Loss"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "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": [
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "plt.figure()\n",
    "plt.title('model accuracy')\n",
    "plt.ylabel('accuracy')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train','test'], loc='upper left')\n",
    "\n",
    "plt.plot(reviews_model.history['acc'])\n",
    "plt.plot(reviews_model.history['val_acc'])\n",
    "\n",
    "plt.show()\n",
    "\n",
    "plt.figure()\n",
    "plt.title('model loss')\n",
    "plt.ylabel('loss')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train','test'], loc='upper left')\n",
    "\n",
    "plt.plot(reviews_model.history['loss'])\n",
    "plt.plot(reviews_model.history['val_loss'])\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Save the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pickle\n",
    "\n",
    "filename = 'model_NN.sav'\n",
    "pickle.dump(model, open(filename, 'wb'))"
   ]
  }
 ],
 "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": "3.7.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}