Sentiment_glove.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "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": "3.6.7"
    },
    "colab": {
      "name": "Sentiment_glove.ipynb",
      "provenance": [],
      "collapsed_sections": []
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "r6ASrmH-iwFI"
      },
      "source": [
        "## **Sentiment analysis on women's clothes reviews by using Glove embeddings and LSTM.**\n",
        "\n",
        "The aim of project is sentiment analysis of the womens's clothes reviews to get a model to predict if the review  is positive or negative. In this approach we used a pre-trained Glove word embeddings and Bidirectional LSTM Neural Network. \n",
        "\n",
        "The Bidirectional LSTMs are an extension of traditional LSTMs and used to ensure that information is passed backward and forward. \n",
        "They may be used on sequence classification problems.\n",
        "The idea of Bidirectional Recurrent Neural Networks (RNNs) is  involves duplicating the first recurrent layer in the network so that there are now two layers side-by-side. Next providing  the input sequence as input to the first layer and providing a reversed copy of the input sequence to the second."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "_T72QHczqZ3V"
      },
      "source": [
        "### **Importing The required libraries**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xZC7zkrEiwFX"
      },
      "source": [
        "import pandas as pd\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "%matplotlib inline\n",
        "import seaborn as sns\n",
        "import os\n",
        "import zipfile\n",
        "\n",
        "import tensorflow as tf\n",
        "from tensorflow.keras.preprocessing.text import Tokenizer\n",
        "from tensorflow.keras.preprocessing.sequence import pad_sequences\n",
        "from tensorflow.keras.layers import Embedding, LSTM, Dense, Bidirectional, Dropout\n",
        "from tensorflow.keras.models import Sequential\n",
        "from tensorflow.keras.callbacks import EarlyStopping\n",
        "\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.metrics import accuracy_score"
      ],
      "execution_count": 2,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "rizPnY27v4rL",
        "outputId": "1bcde289-b703-429b-a402-16b2b25e807c"
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Mounted at /content/drive\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_cHEtV6xvUrg",
        "outputId": "70e1d3c6-a57a-4f56-c6d8-df198e2eda13"
      },
      "source": [
        "cd '/content/drive/My Drive/moje pliki'"
      ],
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/content/drive/My Drive/moje pliki\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "kwCOxmiBzIip"
      },
      "source": [
        "We download the pre-trained word embeddings."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "hT2vsltvzFuu",
        "outputId": "00f48904-26a2-420d-b01c-7f73c9ebda71"
      },
      "source": [
        "!wget --no-check-certificate \\\n",
        "    http://nlp.stanford.edu/data/glove.6B.zip \\\n",
        "    -O /tmp/glove.6B.zip"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "--2021-04-22 13:39:53--  http://nlp.stanford.edu/data/glove.6B.zip\n",
            "Resolving nlp.stanford.edu (nlp.stanford.edu)... 171.64.67.140\n",
            "Connecting to nlp.stanford.edu (nlp.stanford.edu)|171.64.67.140|:80... connected.\n",
            "HTTP request sent, awaiting response... 302 Found\n",
            "Location: https://nlp.stanford.edu/data/glove.6B.zip [following]\n",
            "--2021-04-22 13:39:53--  https://nlp.stanford.edu/data/glove.6B.zip\n",
            "Connecting to nlp.stanford.edu (nlp.stanford.edu)|171.64.67.140|:443... connected.\n",
            "HTTP request sent, awaiting response... 301 Moved Permanently\n",
            "Location: http://downloads.cs.stanford.edu/nlp/data/glove.6B.zip [following]\n",
            "--2021-04-22 13:39:53--  http://downloads.cs.stanford.edu/nlp/data/glove.6B.zip\n",
            "Resolving downloads.cs.stanford.edu (downloads.cs.stanford.edu)... 171.64.64.22\n",
            "Connecting to downloads.cs.stanford.edu (downloads.cs.stanford.edu)|171.64.64.22|:80... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 862182613 (822M) [application/zip]\n",
            "Saving to: ‘/tmp/glove.6B.zip’\n",
            "\n",
            "/tmp/glove.6B.zip   100%[===================>] 822.24M  5.20MB/s    in 2m 40s  \n",
            "\n",
            "2021-04-22 13:42:33 (5.14 MB/s) - ‘/tmp/glove.6B.zip’ saved [862182613/862182613]\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "2ev1KqBu6_Z3"
      },
      "source": [
        "Next we extract them into a temporary folder:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AGqmapTF6-_x"
      },
      "source": [
        "with zipfile.ZipFile('/tmp/glove.6B.zip', 'r') as zip_ref:\n",
        "    zip_ref.extractall('/tmp/glove')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ltfs6MtgyLlp"
      },
      "source": [
        "### **Data preparation**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 206
        },
        "id": "f82_kFeqiwFc",
        "outputId": "0b675bbe-1e96-43e5-e2de-6b298e7c83c2"
      },
      "source": [
        "df = pd.read_csv('reviews_class.csv', header=0,index_col=0)\n",
        "df.head()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "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>Rating</th>\n",
              "      <th>Recommended</th>\n",
              "      <th>Review</th>\n",
              "      <th>Class</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>4</td>\n",
              "      <td>1</td>\n",
              "      <td>Absolutely wonderful - silky and sexy and comf...</td>\n",
              "      <td>Positive</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>5</td>\n",
              "      <td>1</td>\n",
              "      <td>Love this dress!  it's sooo pretty.  i happene...</td>\n",
              "      <td>Positive</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>3</td>\n",
              "      <td>0</td>\n",
              "      <td>I had such high hopes for this dress and reall...</td>\n",
              "      <td>Neutral</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>5</td>\n",
              "      <td>1</td>\n",
              "      <td>I love, love, love this jumpsuit. it's fun, fl...</td>\n",
              "      <td>Positive</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>5</td>\n",
              "      <td>1</td>\n",
              "      <td>This shirt is very flattering to all due to th...</td>\n",
              "      <td>Positive</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   Rating  ...     Class\n",
              "0       4  ...  Positive\n",
              "1       5  ...  Positive\n",
              "2       3  ...   Neutral\n",
              "3       5  ...  Positive\n",
              "4       5  ...  Positive\n",
              "\n",
              "[5 rows x 4 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 7
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 206
        },
        "id": "R6PbjGmQiwFg",
        "outputId": "7363c2f8-d050-41ae-a860-356db3458516"
      },
      "source": [
        "df = df[['Review', 'Recommended']]\n",
        "df.head()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "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>Review</th>\n",
              "      <th>Recommended</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>Absolutely wonderful - silky and sexy and comf...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>Love this dress!  it's sooo pretty.  i happene...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>I had such high hopes for this dress and reall...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>I love, love, love this jumpsuit. it's fun, fl...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>This shirt is very flattering to all due to th...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                                              Review  Recommended\n",
              "0  Absolutely wonderful - silky and sexy and comf...            1\n",
              "1  Love this dress!  it's sooo pretty.  i happene...            1\n",
              "2  I had such high hopes for this dress and reall...            0\n",
              "3  I love, love, love this jumpsuit. it's fun, fl...            1\n",
              "4  This shirt is very flattering to all due to th...            1"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 8
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 206
        },
        "id": "atvqOyVfww-a",
        "outputId": "c2f23894-64fe-4ec9-c739-918c84db3a7d"
      },
      "source": [
        "text = df.rename(columns={\"Recommended\": \"Sentiment\"})\n",
        "text.head()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "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>Review</th>\n",
              "      <th>Sentiment</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>Absolutely wonderful - silky and sexy and comf...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>Love this dress!  it's sooo pretty.  i happene...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>I had such high hopes for this dress and reall...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>I love, love, love this jumpsuit. it's fun, fl...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>This shirt is very flattering to all due to th...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                                              Review  Sentiment\n",
              "0  Absolutely wonderful - silky and sexy and comf...          1\n",
              "1  Love this dress!  it's sooo pretty.  i happene...          1\n",
              "2  I had such high hopes for this dress and reall...          0\n",
              "3  I love, love, love this jumpsuit. it's fun, fl...          1\n",
              "4  This shirt is very flattering to all due to th...          1"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 9
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_1BLTZ3WiwFj",
        "outputId": "5bb24a68-4c75-4207-e5a3-1975beaa3aef"
      },
      "source": [
        "text.shape"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(22641, 2)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 10
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "12UIk6IRiwFk",
        "outputId": "815b31a0-7939-44a5-e980-970db194c92e"
      },
      "source": [
        "text.info()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "<class 'pandas.core.frame.DataFrame'>\n",
            "Int64Index: 22641 entries, 0 to 23485\n",
            "Data columns (total 2 columns):\n",
            " #   Column     Non-Null Count  Dtype \n",
            "---  ------     --------------  ----- \n",
            " 0   Review     22641 non-null  object\n",
            " 1   Sentiment  22641 non-null  int64 \n",
            "dtypes: int64(1), object(1)\n",
            "memory usage: 530.6+ KB\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "qS1U-rxHiwFo",
        "outputId": "7f4e48b1-b6ca-4a10-83c4-731e4fc333d0"
      },
      "source": [
        "text['Sentiment'].value_counts()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "1    18540\n",
              "0     4101\n",
              "Name: Sentiment, dtype: int64"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 13
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "JlYeEKw82G3a"
      },
      "source": [
        "Spliting the data into train and test sets:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kIsK2H1HiwF5"
      },
      "source": [
        "X = text['Review']\n",
        "y = text['Sentiment']\n",
        "\n",
        "X_train_sentences, X_test_sentences, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state = 0)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "IyZV47F84VMT"
      },
      "source": [
        "**Tokenizing the words:**\n",
        "\n",
        "Cause our data is text data, it has to be converted into numerical form since that’s what the deep learning model expects. \n",
        "To convert sentences into numerical representations we use `Tokenizer`, which removes punctuation marks and special characters and converts the sentence to lowercase letter. "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hMhxzpvm2IJr"
      },
      "source": [
        "vocab_size = 5000\n",
        "embedding_dim = 16\n",
        "max_len = 100\n",
        "trunction_type='post'\n",
        "padding_type='post'\n",
        "oov_token = \"<OOV>\""
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NR0YAZmt4W3M"
      },
      "source": [
        "tokenize = Tokenizer(num_words = vocab_size, oov_token=oov_token)\n",
        "tokenize.fit_on_texts(X_train_sentences)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "yNXQ219Y4vQk"
      },
      "source": [
        "The word index show us how words are mapped into numbers:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xqQ7HRjH4XAi"
      },
      "source": [
        "word_index = tokenize.word_index"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "V4C6Eizy0Kv_"
      },
      "source": [
        "By use `texts_to_sequences` from the tokenizer we convert the words into sequences so that a complete sequence of numbers can represent every sentence. "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5Nt-Y4_T-WNv"
      },
      "source": [
        "X_train = tokenize.texts_to_sequences(X_train_sentences)\n",
        "X_test = tokenize.texts_to_sequences(X_test_sentences)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "8Ffr0x9W-3vA",
        "outputId": "2137b0be-4515-4af8-b165-65a4914eadf6"
      },
      "source": [
        "X_train[9:11]"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[[197,\n",
              "  405,\n",
              "  63,\n",
              "  45,\n",
              "  35,\n",
              "  554,\n",
              "  964,\n",
              "  837,\n",
              "  263,\n",
              "  53,\n",
              "  23,\n",
              "  81,\n",
              "  3,\n",
              "  72,\n",
              "  2,\n",
              "  47,\n",
              "  4,\n",
              "  3,\n",
              "  39,\n",
              "  1206],\n",
              " [79, 53]]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 31
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3VLOIF6Q65su"
      },
      "source": [
        "Now we add padding to our data by using `pad_sequences`function to make it uniform."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lcQnbGpG_De0"
      },
      "source": [
        "X_train_pad = pad_sequences(X_train,maxlen=max_len, padding=padding_type, \n",
        "                            truncating=trunction_type)\n",
        "\n",
        "X_test_pad = pad_sequences(X_test,maxlen=max_len, padding=padding_type, \n",
        "                           truncating=trunction_type)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "yRsgtzu6iwGG"
      },
      "source": [
        "### **GloVe Embeddings**\n",
        "We use pre-trained word embeddings to create own embedding layer. We loading the Glove embeddings and append them to a dictionary. \n",
        "\n",
        "Next we use this dictionary to create an embedding matrix for each word in the training set. "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "WcVCWQDliwGH",
        "outputId": "9d97f098-a589-49ed-ee57-4f7e639f5153"
      },
      "source": [
        "embeddings_index = {}\n",
        "f = open('/tmp/glove/glove.6B.100d.txt')\n",
        "for line in f:\n",
        "    values = line.split()\n",
        "    word = values[0]\n",
        "    try:\n",
        "       coefs = np.asarray(values[1:], dtype='float32')\n",
        "       embeddings_index[word] = coefs\n",
        "    except ValueError:\n",
        "       pass\n",
        "f.close()\n",
        "\n",
        "print('Found %s word vectors.' % len(embeddings_index))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Found 400000 word vectors.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Sm6OZObxiwGH"
      },
      "source": [
        "num_words = len(word_index) + 1\n",
        "\n",
        "embedding_matrix = np.zeros((num_words, max_len))\n",
        "for word, i in word_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"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "BYAoj9mfDOdL"
      },
      "source": [
        "Finally we create the embedding layer:\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "oSZ0fmgBiwGH"
      },
      "source": [
        "embedding_layer = Embedding(num_words,\n",
        "                            max_len,\n",
        "                            weights=[embedding_matrix],\n",
        "                            input_length=max_len,\n",
        "                            trainable=False)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "bdcAhZqiDf8X"
      },
      "source": [
        "### **Build the model**\n",
        "We create a model using embedding layer and Bidirectional LSTM layers. Bidirectional LSTMs are supported in Keras via the Bidirectional layer wrapper and it  takes a recurrent layer (e.g. the first LSTM layer) as an argument."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "tGZxZBh6iwGJ",
        "outputId": "592abdd2-f59a-4e71-ba87-ee41b451c23e"
      },
      "source": [
        "model = Sequential([\n",
        "    embedding_layer,\n",
        "    Bidirectional(LSTM(embedding_dim, return_sequences=True)), \n",
        "    Bidirectional(LSTM(embedding_dim)),\n",
        "    Dense(6, activation='relu'),\n",
        "    Dense(1, activation='sigmoid')\n",
        "])\n",
        "\n",
        "model.compile(loss='binary_crossentropy',optimizer='adam',metrics=['accuracy'])\n",
        "\n",
        "model.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "embedding (Embedding)        (None, 100, 100)          1343700   \n",
            "_________________________________________________________________\n",
            "bidirectional (Bidirectional (None, 100, 32)           14976     \n",
            "_________________________________________________________________\n",
            "bidirectional_1 (Bidirection (None, 32)                6272      \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 6)                 198       \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 1)                 7         \n",
            "=================================================================\n",
            "Total params: 1,365,153\n",
            "Trainable params: 21,453\n",
            "Non-trainable params: 1,343,700\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "jyG1Kq8G78iz"
      },
      "source": [
        "We train the model for 10 epoch:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "HD5QDScXDpoT",
        "outputId": "e3645ca5-7a6e-430a-a4f4-cc23ce33b1ab"
      },
      "source": [
        "history = model.fit(X_train_pad, y_train,\n",
        "                    batch_size=32,\n",
        "                    epochs=10,\n",
        "                    validation_data=(X_test_pad, y_test),\n",
        "                    verbose=1)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/10\n",
            "566/566 [==============================] - 47s 76ms/step - loss: 0.4598 - accuracy: 0.8179 - val_loss: 0.4419 - val_accuracy: 0.8075\n",
            "Epoch 2/10\n",
            "566/566 [==============================] - 42s 73ms/step - loss: 0.3995 - accuracy: 0.8302 - val_loss: 0.3954 - val_accuracy: 0.8337\n",
            "Epoch 3/10\n",
            "566/566 [==============================] - 41s 72ms/step - loss: 0.3542 - accuracy: 0.8404 - val_loss: 0.3663 - val_accuracy: 0.8267\n",
            "Epoch 4/10\n",
            "566/566 [==============================] - 41s 73ms/step - loss: 0.3464 - accuracy: 0.8447 - val_loss: 0.3827 - val_accuracy: 0.8064\n",
            "Epoch 5/10\n",
            "566/566 [==============================] - 42s 74ms/step - loss: 0.3483 - accuracy: 0.8319 - val_loss: 0.3186 - val_accuracy: 0.8613\n",
            "Epoch 6/10\n",
            "566/566 [==============================] - 42s 75ms/step - loss: 0.3023 - accuracy: 0.8674 - val_loss: 0.3050 - val_accuracy: 0.8649\n",
            "Epoch 7/10\n",
            "566/566 [==============================] - 43s 76ms/step - loss: 0.2817 - accuracy: 0.8774 - val_loss: 0.2900 - val_accuracy: 0.8744\n",
            "Epoch 8/10\n",
            "566/566 [==============================] - 43s 75ms/step - loss: 0.2676 - accuracy: 0.8862 - val_loss: 0.2808 - val_accuracy: 0.8761\n",
            "Epoch 9/10\n",
            "566/566 [==============================] - 42s 75ms/step - loss: 0.2502 - accuracy: 0.8911 - val_loss: 0.2881 - val_accuracy: 0.8739\n",
            "Epoch 10/10\n",
            "566/566 [==============================] - 41s 73ms/step - loss: 0.2429 - accuracy: 0.8956 - val_loss: 0.2843 - val_accuracy: 0.8803\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0G4YjehB6Dur"
      },
      "source": [
        "Predictions on a test set:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "mZCgDZQtDpuZ",
        "outputId": "a76709fd-2c95-49b0-9194-ccbc4c122075"
      },
      "source": [
        "loss, accuracy = model.evaluate(X_test_pad,y_test)\n",
        "print('Test accuracy :', accuracy)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "142/142 [==============================] - 3s 21ms/step - loss: 0.2843 - accuracy: 0.8803\n",
            "Test accuracy : 0.8803268074989319\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 544
        },
        "id": "szrloPlxDpyf",
        "outputId": "38b7340d-247b-42a1-e07d-634dfb3134e6"
      },
      "source": [
        "def plot_graphs(history, string):\n",
        "  plt.plot(history.history[string])\n",
        "  plt.plot(history.history['val_'+string])\n",
        "  plt.xlabel(\"Epochs\")\n",
        "  plt.ylabel(string)\n",
        "  plt.legend([string, 'val_'+string])\n",
        "  plt.show()\n",
        "  \n",
        "plot_graphs(history, \"accuracy\")\n",
        "plot_graphs(history, \"loss\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ULN3VM7NexNu"
      },
      "source": [
        "model.save(\"sentiment_model.h5\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "URf-oephHjQU"
      },
      "source": [
        "**Model 2**\n",
        "\n",
        "We also test a second model with a Dropout layer for decrease overfitting and  we increase number of epochs to 20."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "gI3cAdVrf_y-",
        "outputId": "7c3a97bc-382a-47d8-e5ee-74b2a7ac4a08"
      },
      "source": [
        "model2 = Sequential([\n",
        "    embedding_layer,\n",
        "    Bidirectional(LSTM(64, return_sequences=True)), \n",
        "    Bidirectional(LSTM(32)),\n",
        "    Dense(16, activation='relu'),\n",
        "    Dropout(0.5),\n",
        "    Dense(1, activation='sigmoid')\n",
        "])\n",
        "\n",
        "optimizer = tf.keras.optimizers.Adam(lr=1e-4)\n",
        "model2.compile(loss='binary_crossentropy', \n",
        "               optimizer=optimizer,\n",
        "               metrics=['accuracy'])\n",
        "\n",
        "model2.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential_6\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "embedding (Embedding)        (None, 100, 100)          1343700   \n",
            "_________________________________________________________________\n",
            "bidirectional_14 (Bidirectio (None, 100, 128)          84480     \n",
            "_________________________________________________________________\n",
            "bidirectional_15 (Bidirectio (None, 64)                41216     \n",
            "_________________________________________________________________\n",
            "dense_13 (Dense)             (None, 16)                1040      \n",
            "_________________________________________________________________\n",
            "dropout_4 (Dropout)          (None, 16)                0         \n",
            "_________________________________________________________________\n",
            "dense_14 (Dense)             (None, 1)                 17        \n",
            "=================================================================\n",
            "Total params: 1,470,453\n",
            "Trainable params: 126,753\n",
            "Non-trainable params: 1,343,700\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7qJwNUOQJOcC"
      },
      "source": [
        "callbacks = EarlyStopping(monitor = 'val_accuracy', \n",
        "                          mode = 'max', \n",
        "                          patience = 10,\n",
        "                          verbose = 1)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "uztzED5_f_u8",
        "outputId": "4f7faac4-5ccd-4ccb-8896-71c01ad88687"
      },
      "source": [
        "history2 = model2.fit(X_train_pad, y_train,\n",
        "                    batch_size=32,\n",
        "                    epochs=20,\n",
        "                    validation_data=(X_test_pad, y_test),\n",
        "                    verbose=1,\n",
        "                    callbacks=callbacks)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/20\n",
            "566/566 [==============================] - 92s 152ms/step - loss: 0.5288 - accuracy: 0.7981 - val_loss: 0.4305 - val_accuracy: 0.8064\n",
            "Epoch 2/20\n",
            "566/566 [==============================] - 85s 150ms/step - loss: 0.4186 - accuracy: 0.8241 - val_loss: 0.3872 - val_accuracy: 0.8159\n",
            "Epoch 3/20\n",
            "566/566 [==============================] - 85s 150ms/step - loss: 0.3853 - accuracy: 0.8362 - val_loss: 0.3736 - val_accuracy: 0.8359\n",
            "Epoch 4/20\n",
            "566/566 [==============================] - 85s 150ms/step - loss: 0.3608 - accuracy: 0.8461 - val_loss: 0.3710 - val_accuracy: 0.8306\n",
            "Epoch 5/20\n",
            "566/566 [==============================] - 85s 150ms/step - loss: 0.3556 - accuracy: 0.8499 - val_loss: 0.3461 - val_accuracy: 0.8488\n",
            "Epoch 6/20\n",
            "566/566 [==============================] - 85s 149ms/step - loss: 0.3325 - accuracy: 0.8602 - val_loss: 0.3241 - val_accuracy: 0.8589\n",
            "Epoch 7/20\n",
            "566/566 [==============================] - 85s 150ms/step - loss: 0.3204 - accuracy: 0.8626 - val_loss: 0.3182 - val_accuracy: 0.8613\n",
            "Epoch 8/20\n",
            "566/566 [==============================] - 85s 150ms/step - loss: 0.3179 - accuracy: 0.8642 - val_loss: 0.3156 - val_accuracy: 0.8616\n",
            "Epoch 9/20\n",
            "566/566 [==============================] - 85s 150ms/step - loss: 0.3040 - accuracy: 0.8740 - val_loss: 0.3094 - val_accuracy: 0.8651\n",
            "Epoch 10/20\n",
            "566/566 [==============================] - 85s 151ms/step - loss: 0.2948 - accuracy: 0.8754 - val_loss: 0.3162 - val_accuracy: 0.8662\n",
            "Epoch 11/20\n",
            "566/566 [==============================] - 85s 150ms/step - loss: 0.3031 - accuracy: 0.8713 - val_loss: 0.3011 - val_accuracy: 0.8664\n",
            "Epoch 12/20\n",
            "566/566 [==============================] - 85s 150ms/step - loss: 0.2882 - accuracy: 0.8741 - val_loss: 0.2977 - val_accuracy: 0.8691\n",
            "Epoch 13/20\n",
            "566/566 [==============================] - 85s 150ms/step - loss: 0.2766 - accuracy: 0.8824 - val_loss: 0.3008 - val_accuracy: 0.8666\n",
            "Epoch 14/20\n",
            "566/566 [==============================] - 85s 150ms/step - loss: 0.2804 - accuracy: 0.8819 - val_loss: 0.2917 - val_accuracy: 0.8704\n",
            "Epoch 15/20\n",
            "566/566 [==============================] - 85s 150ms/step - loss: 0.2697 - accuracy: 0.8866 - val_loss: 0.3006 - val_accuracy: 0.8691\n",
            "Epoch 16/20\n",
            "566/566 [==============================] - 85s 150ms/step - loss: 0.2721 - accuracy: 0.8833 - val_loss: 0.3165 - val_accuracy: 0.8638\n",
            "Epoch 17/20\n",
            "566/566 [==============================] - 82s 144ms/step - loss: 0.2685 - accuracy: 0.8872 - val_loss: 0.2926 - val_accuracy: 0.8741\n",
            "Epoch 18/20\n",
            "566/566 [==============================] - 79s 139ms/step - loss: 0.2511 - accuracy: 0.8910 - val_loss: 0.2867 - val_accuracy: 0.8724\n",
            "Epoch 19/20\n",
            "566/566 [==============================] - 79s 140ms/step - loss: 0.2682 - accuracy: 0.8896 - val_loss: 0.2865 - val_accuracy: 0.8719\n",
            "Epoch 20/20\n",
            "566/566 [==============================] - 78s 137ms/step - loss: 0.2434 - accuracy: 0.8948 - val_loss: 0.2887 - val_accuracy: 0.8752\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "qzNXAOga7XHR"
      },
      "source": [
        "Predictions:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "JLeI5Apzf_o8",
        "outputId": "a1ae6d19-d408-4879-b3f7-064e6283e35f"
      },
      "source": [
        "loss, accuracy = model2.evaluate(X_test_pad,y_test)\n",
        "print('Test accuracy :', accuracy)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "142/142 [==============================] - 5s 37ms/step - loss: 0.2887 - accuracy: 0.8752\n",
            "Test accuracy : 0.875248372554779\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 541
        },
        "id": "HxYIdfwlnfNd",
        "outputId": "27d9f17a-73c6-42a7-af8a-4cd4cc7ea08e"
      },
      "source": [
        "plot_graphs(history2, \"accuracy\")\n",
        "plot_graphs(history2, \"loss\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "VhEgsIGa8DKt"
      },
      "source": [
        "### **Summary**\n",
        "\n",
        "In this attempt we used a pre-trained Glove word embeddings with Bidirectional LSTM Neural Network to resolve our classification problem. After trained two models we achieved an accuracy on the test set 88 % and it is a good result in comparison to previous Machine Learning models that we have used (e.g. Logistic Regression)."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jR1k2IoP8HZt"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}