hotel_review_bert_RMDL.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "hotel_review_bert_RMDL.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "TPU"
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "X_GirHZVJQGz"
      },
      "source": [
        "import numpy as np\n",
        "import pandas as pd\n",
        "import matplotlib.pyplot as plt\n",
        "from sklearn.preprocessing import LabelEncoder\n",
        "from nltk.corpus import stopwords\n",
        "import tensorflow as tf\n",
        "from tensorflow.keras.preprocessing.text import Tokenizer"
      ],
      "execution_count": 61,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "s0v4HsCxJk_I",
        "outputId": "bf727426-9383-456f-8eae-54ef3a2c21e6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 36
        }
      },
      "source": [
        "tf.__version__"
      ],
      "execution_count": 62,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            },
            "text/plain": [
              "'2.3.0'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 62
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "sF3vhIuWpvSh",
        "outputId": "e27ee557-1dcf-45b5-9884-5572bb10f2da",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ],
      "execution_count": 63,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount(\"/content/drive\", force_remount=True).\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XlYwGQ6GJnDL"
      },
      "source": [
        "hotel_review = pd.read_csv('/content/drive/My Drive/data/Hotel_Reviews.csv')"
      ],
      "execution_count": 64,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "95qf_WgHJwA9",
        "outputId": "31579556-d2e6-45f4-c830-1857e32aba7b",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 695
        }
      },
      "source": [
        "hotel_review.head()"
      ],
      "execution_count": 65,
      "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>Hotel_Address</th>\n",
              "      <th>Additional_Number_of_Scoring</th>\n",
              "      <th>Review_Date</th>\n",
              "      <th>Average_Score</th>\n",
              "      <th>Hotel_Name</th>\n",
              "      <th>Reviewer_Nationality</th>\n",
              "      <th>Negative_Review</th>\n",
              "      <th>Review_Total_Negative_Word_Counts</th>\n",
              "      <th>Total_Number_of_Reviews</th>\n",
              "      <th>Positive_Review</th>\n",
              "      <th>Review_Total_Positive_Word_Counts</th>\n",
              "      <th>Total_Number_of_Reviews_Reviewer_Has_Given</th>\n",
              "      <th>Reviewer_Score</th>\n",
              "      <th>Tags</th>\n",
              "      <th>days_since_review</th>\n",
              "      <th>lat</th>\n",
              "      <th>lng</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>s Gravesandestraat 55 Oost 1092 AA Amsterdam ...</td>\n",
              "      <td>194</td>\n",
              "      <td>8/3/2017</td>\n",
              "      <td>7.7</td>\n",
              "      <td>Hotel Arena</td>\n",
              "      <td>Russia</td>\n",
              "      <td>I am so angry that i made this post available...</td>\n",
              "      <td>397</td>\n",
              "      <td>1403</td>\n",
              "      <td>Only the park outside of the hotel was beauti...</td>\n",
              "      <td>11</td>\n",
              "      <td>7</td>\n",
              "      <td>2.9</td>\n",
              "      <td>[' Leisure trip ', ' Couple ', ' Duplex Double...</td>\n",
              "      <td>0 days</td>\n",
              "      <td>52.360576</td>\n",
              "      <td>4.915968</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>s Gravesandestraat 55 Oost 1092 AA Amsterdam ...</td>\n",
              "      <td>194</td>\n",
              "      <td>8/3/2017</td>\n",
              "      <td>7.7</td>\n",
              "      <td>Hotel Arena</td>\n",
              "      <td>Ireland</td>\n",
              "      <td>No Negative</td>\n",
              "      <td>0</td>\n",
              "      <td>1403</td>\n",
              "      <td>No real complaints the hotel was great great ...</td>\n",
              "      <td>105</td>\n",
              "      <td>7</td>\n",
              "      <td>7.5</td>\n",
              "      <td>[' Leisure trip ', ' Couple ', ' Duplex Double...</td>\n",
              "      <td>0 days</td>\n",
              "      <td>52.360576</td>\n",
              "      <td>4.915968</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>s Gravesandestraat 55 Oost 1092 AA Amsterdam ...</td>\n",
              "      <td>194</td>\n",
              "      <td>7/31/2017</td>\n",
              "      <td>7.7</td>\n",
              "      <td>Hotel Arena</td>\n",
              "      <td>Australia</td>\n",
              "      <td>Rooms are nice but for elderly a bit difficul...</td>\n",
              "      <td>42</td>\n",
              "      <td>1403</td>\n",
              "      <td>Location was good and staff were ok It is cut...</td>\n",
              "      <td>21</td>\n",
              "      <td>9</td>\n",
              "      <td>7.1</td>\n",
              "      <td>[' Leisure trip ', ' Family with young childre...</td>\n",
              "      <td>3 days</td>\n",
              "      <td>52.360576</td>\n",
              "      <td>4.915968</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>s Gravesandestraat 55 Oost 1092 AA Amsterdam ...</td>\n",
              "      <td>194</td>\n",
              "      <td>7/31/2017</td>\n",
              "      <td>7.7</td>\n",
              "      <td>Hotel Arena</td>\n",
              "      <td>United Kingdom</td>\n",
              "      <td>My room was dirty and I was afraid to walk ba...</td>\n",
              "      <td>210</td>\n",
              "      <td>1403</td>\n",
              "      <td>Great location in nice surroundings the bar a...</td>\n",
              "      <td>26</td>\n",
              "      <td>1</td>\n",
              "      <td>3.8</td>\n",
              "      <td>[' Leisure trip ', ' Solo traveler ', ' Duplex...</td>\n",
              "      <td>3 days</td>\n",
              "      <td>52.360576</td>\n",
              "      <td>4.915968</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>s Gravesandestraat 55 Oost 1092 AA Amsterdam ...</td>\n",
              "      <td>194</td>\n",
              "      <td>7/24/2017</td>\n",
              "      <td>7.7</td>\n",
              "      <td>Hotel Arena</td>\n",
              "      <td>New Zealand</td>\n",
              "      <td>You When I booked with your company on line y...</td>\n",
              "      <td>140</td>\n",
              "      <td>1403</td>\n",
              "      <td>Amazing location and building Romantic setting</td>\n",
              "      <td>8</td>\n",
              "      <td>3</td>\n",
              "      <td>6.7</td>\n",
              "      <td>[' Leisure trip ', ' Couple ', ' Suite ', ' St...</td>\n",
              "      <td>10 days</td>\n",
              "      <td>52.360576</td>\n",
              "      <td>4.915968</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                                       Hotel_Address  ...       lng\n",
              "0   s Gravesandestraat 55 Oost 1092 AA Amsterdam ...  ...  4.915968\n",
              "1   s Gravesandestraat 55 Oost 1092 AA Amsterdam ...  ...  4.915968\n",
              "2   s Gravesandestraat 55 Oost 1092 AA Amsterdam ...  ...  4.915968\n",
              "3   s Gravesandestraat 55 Oost 1092 AA Amsterdam ...  ...  4.915968\n",
              "4   s Gravesandestraat 55 Oost 1092 AA Amsterdam ...  ...  4.915968\n",
              "\n",
              "[5 rows x 17 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 65
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "W4-yBpI6JzU0",
        "outputId": "1682f12e-3484-44bc-d55e-aabc4a45b5e9",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "hotel_review.shape"
      ],
      "execution_count": 66,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(515738, 17)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 66
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "dLpI4KinJ1R_"
      },
      "source": [
        "positive_review = hotel_review.loc[:,['Positive_Review']]\n",
        "negative_review = hotel_review.loc[:,['Negative_Review']]"
      ],
      "execution_count": 67,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mDTA3C9fJ3A8",
        "outputId": "b271a2a4-12f1-45bb-dd2b-992d1ed165fe",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 316
        }
      },
      "source": [
        "options = ['No Positive', 'No Negative']\n",
        "selected_positive_review = positive_review.loc[~(positive_review['Positive_Review'] == \"No Positive\"), :]\n",
        "selected_negative_review = negative_review.loc[~(negative_review['Negative_Review'] == \"No Negative\"), :]\n",
        "selected_positive_review['is_positive'] = 1\n",
        "selected_positive_review.columns = ['review', 'is_positive']\n",
        "selected_positive_review.head()"
      ],
      "execution_count": 68,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:4: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  after removing the cwd from sys.path.\n"
          ],
          "name": "stderr"
        },
        {
          "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>is_positive</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>Only the park outside of the hotel was beauti...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>No real complaints the hotel was great great ...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>Location was good and staff were ok It is cut...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>Great location in nice surroundings the bar a...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>Amazing location and building Romantic setting</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                                              review  is_positive\n",
              "0   Only the park outside of the hotel was beauti...            1\n",
              "1   No real complaints the hotel was great great ...            1\n",
              "2   Location was good and staff were ok It is cut...            1\n",
              "3   Great location in nice surroundings the bar a...            1\n",
              "4    Amazing location and building Romantic setting             1"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 68
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PZvzv9FnJ5B6",
        "outputId": "a3fe9544-300e-478f-c4bb-877ce5d77521",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "selected_negative_review['is_postive'] = 0\n",
        "selected_negative_review.columns = ['review', 'is_positive']"
      ],
      "execution_count": 69,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:1: SettingWithCopyWarning: \n",
            "A value is trying to be set on a copy of a slice from a DataFrame.\n",
            "Try using .loc[row_indexer,col_indexer] = value instead\n",
            "\n",
            "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
            "  \"\"\"Entry point for launching an IPython kernel.\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "COcr_2wBJ9uO"
      },
      "source": [
        "selected_positive_review = selected_positive_review[:30000]\n",
        "selected_negative_review = selected_negative_review[:30000]"
      ],
      "execution_count": 70,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "A7MRGZ8SJ_i3",
        "outputId": "45768b9f-5706-4fd2-b5fd-90a64af4ef97",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "print(selected_negative_review.shape)\n",
        "print(selected_positive_review.shape)\n"
      ],
      "execution_count": 71,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(30000, 2)\n",
            "(30000, 2)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MJ2v-txGKBAn"
      },
      "source": [
        "reviews = selected_positive_review.append(selected_negative_review)"
      ],
      "execution_count": 72,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fpBQtxLQKDp_",
        "outputId": "716b806f-bd23-474f-c103-c0678d721d4f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 363
        }
      },
      "source": [
        "reviews.head(n=10)"
      ],
      "execution_count": 73,
      "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>is_positive</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>Only the park outside of the hotel was beauti...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>No real complaints the hotel was great great ...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>Location was good and staff were ok It is cut...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>Great location in nice surroundings the bar a...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>Amazing location and building Romantic setting</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5</th>\n",
              "      <td>Good restaurant with modern design great chil...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6</th>\n",
              "      <td>The room is spacious and bright The hotel is ...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>7</th>\n",
              "      <td>Good location Set in a lovely park friendly s...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>9</th>\n",
              "      <td>The room was big enough and the bed is good T...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>10</th>\n",
              "      <td>Rooms were stunningly decorated and really sp...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                                               review  is_positive\n",
              "0    Only the park outside of the hotel was beauti...            1\n",
              "1    No real complaints the hotel was great great ...            1\n",
              "2    Location was good and staff were ok It is cut...            1\n",
              "3    Great location in nice surroundings the bar a...            1\n",
              "4     Amazing location and building Romantic setting             1\n",
              "5    Good restaurant with modern design great chil...            1\n",
              "6    The room is spacious and bright The hotel is ...            1\n",
              "7    Good location Set in a lovely park friendly s...            1\n",
              "9    The room was big enough and the bed is good T...            1\n",
              "10   Rooms were stunningly decorated and really sp...            1"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 73
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PgVQuV_uKFN3",
        "outputId": "fc9c9dc1-8e47-4722-f1a2-628bc23cb645",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "import nltk\n",
        "nltk.download('stopwords')\n",
        "nltk.download('wordnet')\n",
        "nltk.download('averaged_perceptron_tagger')\n",
        "from nltk.corpus import wordnet\n",
        "import string\n",
        "from nltk import pos_tag\n",
        "from nltk.corpus import stopwords\n",
        "from nltk.tokenize import WhitespaceTokenizer\n",
        "from nltk.stem import WordNetLemmatizer"
      ],
      "execution_count": 74,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[nltk_data] Downloading package stopwords to /root/nltk_data...\n",
            "[nltk_data]   Package stopwords is already up-to-date!\n",
            "[nltk_data] Downloading package wordnet to /root/nltk_data...\n",
            "[nltk_data]   Package wordnet is already up-to-date!\n",
            "[nltk_data] Downloading package averaged_perceptron_tagger to\n",
            "[nltk_data]     /root/nltk_data...\n",
            "[nltk_data]   Package averaged_perceptron_tagger is already up-to-\n",
            "[nltk_data]       date!\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "D8iE4D8e6xjy"
      },
      "source": [
        "# New section"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gjeHdSW8KHPh"
      },
      "source": [
        "def get_wordnet_pos(pos_tag):\n",
        "    if pos_tag.startswith('J'):\n",
        "        return wordnet.ADJ\n",
        "    elif pos_tag.startswith('V'):\n",
        "        return wordnet.VERB\n",
        "    elif pos_tag.startswith('N'):\n",
        "        return wordnet.NOUN\n",
        "    elif pos_tag.startswith('R'):\n",
        "        return wordnet.ADV\n",
        "    else:\n",
        "        return wordnet.NOUN\n",
        "\n",
        "def clean_text(text):\n",
        "    # lower text\n",
        "    text = text.lower()\n",
        "    # tokenize text and remove puncutation\n",
        "    text = [word.strip(string.punctuation) for word in text.split(\" \")]\n",
        "    # remove words that contain numbers\n",
        "    text = [word for word in text if not any(c.isdigit() for c in word)]\n",
        "    # remove stop words\n",
        "    stop = stopwords.words('english')\n",
        "    text = [x for x in text if x not in stop]\n",
        "    # remove empty tokens\n",
        "    text = [t for t in text if len(t) > 0]\n",
        "    # pos tag text\n",
        "    pos_tags = pos_tag(text)\n",
        "    # lemmatize text\n",
        "    text = [WordNetLemmatizer().lemmatize(t[0], get_wordnet_pos(t[1])) for t in pos_tags]\n",
        "    # remove words with only one letter\n",
        "    text = [t for t in text if len(t) > 1]\n",
        "    # join all\n",
        "    text = \" \".join(text)\n",
        "    return(text)"
      ],
      "execution_count": 75,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nQ88VDoxKJYL"
      },
      "source": [
        "# clean text data\n",
        "reviews[\"review\"] = reviews[\"review\"].apply(lambda x: clean_text(x))"
      ],
      "execution_count": 76,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "j2HsozdyOuDH",
        "outputId": "a35cd9f3-e91c-48e2-8b4a-c777ec522587",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "reviews_text = reviews['review'].tolist()\n",
        "labels = reviews['is_positive'].tolist()\n",
        "\n",
        "enc = LabelEncoder()\n",
        "encoded_labels = enc.fit_transform(labels)\n",
        "\n",
        "\n",
        "#TOKENIZE\n",
        "from nltk.tokenize import sent_tokenize\n",
        "from nltk.tokenize import word_tokenize\n",
        "from keras.preprocessing.text import Tokenizer\n",
        "import nltk\n",
        "nltk.download('punkt')\n",
        "complete_reviews=[] \n",
        "for review in reviews['review']:\n",
        "  for sent in sent_tokenize(review):\n",
        "    complete_reviews.append(sent)\n",
        "print('complete_reviews done') \n",
        "final_corpus = [sentence for sentence in complete_reviews if sentence.strip() !='']\n",
        "print('final corpus done')\n",
        "word_punctuation_tokenizer = nltk.WordPunctTokenizer()\n",
        "word_tokenized_corpus = [word_punctuation_tokenizer.tokenize(sent) for sent in final_corpus]"
      ],
      "execution_count": 77,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[nltk_data] Downloading package punkt to /root/nltk_data...\n",
            "[nltk_data]   Package punkt is already up-to-date!\n",
            "complete_reviews done\n",
            "final corpus done\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GRTTsdI9Oxfg"
      },
      "source": [
        "# model parameters\n",
        "vocab_size = 28593\n",
        "embedding_dim = 128\n",
        "max_length = 128\n",
        "trunc_type='post'\n",
        "padding_type='post'\n",
        "oov_tok = \"<OOV>\"\n",
        "training_portion=0.7"
      ],
      "execution_count": 78,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-OuyohhwO2vc",
        "outputId": "cebd1e81-1006-4538-a375-f9e94daf8119",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 298
        }
      },
      "source": [
        "import seaborn as sns\n",
        "sns.countplot(x=\"is_positive\",data=reviews)"
      ],
      "execution_count": 79,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7fb1f2f558d0>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 79
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZEAAAEHCAYAAABvHnsJAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAT+ElEQVR4nO3dfaxd1X3m8e+DCTTNGyS4DLVNjBK3I6dtDLHA03QqhrRgqBrTimYganAYFHdUmDaaKAqpoiFDgtSoSVFJUypSXEybwSFJM7iRW9eliCiTADYJBQzDcEuSwRYBF/PaqGRMf/PHWbccmXud42Wfe3253490dPf+7bXXXhshHvbLWSdVhSRJPY6Y7QFIkuYuQ0SS1M0QkSR1M0QkSd0MEUlStyNnewAz7bjjjqulS5fO9jAkaU656667/rGqFu5bn3chsnTpUrZv3z7bw5CkOSXJd6eqeztLktTNEJEkdTNEJEndDBFJUjdDRJLUzRCRJHUbW4gk+ZEkdyb5+yQ7kvz3Vj8pyR1JJpJ8PslRrX50W59o25cO9fXhVn8wyVlD9dWtNpHksnGdiyRpauO8EnkeOKOq3gqsAFYnWQV8Ariqqt4MPAlc3NpfDDzZ6le1diRZDpwPvAVYDfxRkgVJFgCfAc4GlgMXtLaSpBkythCpgefa6ivap4AzgC+2+gbg3La8pq3Ttr8jSVp9Y1U9X1XfBiaAU9tnoqoerqofABtbW0nSDBnrN9bb1cJdwJsZXDX8A/BUVe1tTXYCi9ryIuARgKram+Rp4A2tfvtQt8P7PLJP/bRpxrEOWAdw4oknHtQ5ve2DNxzU/np5uuv3LpztIQDwf6/46dkegg5DJ/63e8fW91gfrFfVC1W1AljM4Mrh347zePsZx7VVtbKqVi5c+JKpXyRJnWbk7ayqegq4Ffh3wDFJJq+AFgO72vIuYAlA2/464Inh+j77TFeXJM2Qcb6dtTDJMW35lcAvAg8wCJPzWrO1wM1teVNbp23/uxr8APwm4Pz29tZJwDLgTmAbsKy97XUUg4fvm8Z1PpKklxrnM5ETgA3tucgRwE1V9ZUk9wMbk3wc+BZwXWt/HfBnSSaAPQxCgarakeQm4H5gL3BJVb0AkORSYAuwAFhfVTvGeD6SpH2MLUSq6h7g5CnqDzN4PrJv/Z+BX5umryuBK6eobwY2H/RgJUld/Ma6JKmbISJJ6maISJK6GSKSpG6GiCSpmyEiSepmiEiSuhkikqRuhogkqZshIknqZohIkroZIpKkboaIJKmbISJJ6maISJK6GSKSpG6GiCSpmyEiSepmiEiSuhkikqRuhogkqZshIknqZohIkroZIpKkboaIJKnb2EIkyZIktya5P8mOJL/d6h9NsivJ3e1zztA+H04ykeTBJGcN1Ve32kSSy4bqJyW5o9U/n+SocZ2PJOmlxnklshf4QFUtB1YBlyRZ3rZdVVUr2mczQNt2PvAWYDXwR0kWJFkAfAY4G1gOXDDUzydaX28GngQuHuP5SJL2MbYQqapHq+qbbflZ4AFg0X52WQNsrKrnq+rbwARwavtMVNXDVfUDYCOwJkmAM4Avtv03AOeO52wkSVOZkWciSZYCJwN3tNKlSe5Jsj7Jsa22CHhkaLedrTZd/Q3AU1W1d5/6VMdfl2R7ku27d+8+BGckSYIZCJEkrwa+BLy/qp4BrgHeBKwAHgU+Ne4xVNW1VbWyqlYuXLhw3IeTpHnjyHF2nuQVDALkc1X1FwBV9djQ9s8CX2mru4AlQ7svbjWmqT8BHJPkyHY1MtxekjQDxvl2VoDrgAeq6veH6icMNfsV4L62vAk4P8nRSU4ClgF3AtuAZe1NrKMYPHzfVFUF3Aqc1/ZfC9w8rvORJL3UOK9E3g68B7g3yd2t9jsM3q5aARTwHeA3AKpqR5KbgPsZvNl1SVW9AJDkUmALsABYX1U7Wn8fAjYm+TjwLQahJUmaIWMLkar6GpApNm3ezz5XAldOUd881X5V9TCDt7ckSbPAb6xLkroZIpKkboaIJKmbISJJ6maISJK6GSKSpG6GiCSpmyEiSepmiEiSuhkikqRuhogkqZshIknqZohIkroZIpKkboaIJKmbISJJ6maISJK6GSKSpG6GiCSpmyEiSepmiEiSuhkikqRuhogkqZshIknqZohIkroZIpKkbmMLkSRLktya5P4kO5L8dqu/PsnWJA+1v8e2epJcnWQiyT1JThnqa21r/1CStUP1tyW5t+1zdZKM63wkSS81ziuRvcAHqmo5sAq4JMly4DLglqpaBtzS1gHOBpa1zzrgGhiEDnA5cBpwKnD5ZPC0Nu8b2m/1GM9HkrSPsYVIVT1aVd9sy88CDwCLgDXAhtZsA3BuW14D3FADtwPHJDkBOAvYWlV7qupJYCuwum17bVXdXlUF3DDUlyRpBszIM5EkS4GTgTuA46vq0bbpe8DxbXkR8MjQbjtbbX/1nVPUpzr+uiTbk2zfvXv3QZ2LJOlFYw+RJK8GvgS8v6qeGd7WriBq3GOoqmuramVVrVy4cOG4DydJ88ZYQyTJKxgEyOeq6i9a+bF2K4r29/FW3wUsGdp9cavtr754irokaYaM8+2sANcBD1TV7w9t2gRMvmG1Frh5qH5he0trFfB0u+21BTgzybHtgfqZwJa27Zkkq9qxLhzqS5I0A44cY99vB94D3Jvk7lb7HeB3gZuSXAx8F3hX27YZOAeYAL4PXARQVXuSfAzY1tpdUVV72vJvAtcDrwT+qn0kSTNkbCFSVV8DpvvexjumaF/AJdP0tR5YP0V9O/BTBzFMSdJB8BvrkqRuhogkqZshIknqZohIkroZIpKkboaIJKmbISJJ6maISJK6GSKSpG6GiCSpmyEiSepmiEiSuo0UIkluGaUmSZpf9juLb5IfAX4UOK79lsfkrLyvZZqfopUkzR8/bCr43wDeD/w4cBcvhsgzwB+OcVySpDlgvyFSVX8A/EGS/1JVn56hMUmS5oiRfpSqqj6d5GeBpcP7VNUNYxqXJGkOGClEkvwZ8CbgbuCFVi7AEJGkeWzUn8ddCSxvP2ErSRIw+vdE7gP+zTgHIkmae0a9EjkOuD/JncDzk8WqeudYRiVJmhNGDZGPjnMQkqS5adS3s24b90AkSXPPqG9nPcvgbSyAo4BXAP9UVa8d18AkSYe/Ua9EXjO5nCTAGmDVuAYlSZobDngW3xr4n8BZ+2uXZH2Sx5PcN1T7aJJdSe5un3OGtn04yUSSB5OcNVRf3WoTSS4bqp+U5I5W/3ySow70XCRJB2fU21m/OrR6BIPvjfzzD9ntegbza+37hcSrquqT+/S/HDgfeAuDebr+NslPtM2fAX4R2AlsS7Kpqu4HPtH62pjkj4GLgWtGOR9J0qEx6ttZvzy0vBf4DoNbWtOqqq8mWTpi/2uAjVX1PPDtJBPAqW3bRFU9DJBkI7AmyQPAGcC7W5sNDN4gM0QkaQaN+kzkokN4zEuTXAhsBz5QVU8ymFb+9qE2O3lxqvlH9qmfBrwBeKqq9k7R/iWSrAPWAZx44omH4hwkSYz+o1SLk3y5PeN4PMmXkizuON41DObgWgE8Cnyqo48DVlXXVtXKqlq5cOHCmTikJM0Loz5Y/1NgE4PnFT8O/GWrHZCqeqyqXqiqfwE+y4u3rHYBS4aaLm616epPAMckOXKfuiRpBo0aIgur6k+ram/7XA8c8P/SJzlhaPVXGMzJBYOAOj/J0UlOApYBdwLbgGXtTayjGDx839QmgrwVOK/tvxa4+UDHI0k6OKM+WH8iya8DN7b1CxhcDUwryY3A6Qx+WncncDlwepIVDL64+B0Gv5xIVe1IchNwP4MH95dU1Qutn0uBLcACYH1V7WiH+BCwMcnHgW8B1414LpKkQ2TUEPlPwKeBqxgEwNeB9+5vh6q6YIrytP+hr6orgSunqG8GNk9Rf5gXb4dJkmbBqCFyBbC2vUlFktcDn2QQLpKkeWrUZyI/MxkgAFW1Bzh5PEOSJM0Vo4bIEUmOnVxpVyKjXsVIkl6mRg2CTwHfSPKFtv5rTPH8QpI0v4z6jfUbkmxnMNUIwK+2+askSfPYyLekWmgYHJKkf3XAU8FLkjTJEJEkdTNEJEndDBFJUjdDRJLUzRCRJHUzRCRJ3QwRSVI3Q0SS1M0QkSR1M0QkSd0MEUlSN0NEktTNEJEkdTNEJEndDBFJUjdDRJLUzRCRJHUzRCRJ3QwRSVK3sYVIkvVJHk9y31Dt9Um2Jnmo/T221ZPk6iQTSe5JcsrQPmtb+4eSrB2qvy3JvW2fq5NkXOciSZraOK9ErgdW71O7DLilqpYBt7R1gLOBZe2zDrgGBqEDXA6cBpwKXD4ZPK3N+4b22/dYkqQxG1uIVNVXgT37lNcAG9ryBuDcofoNNXA7cEySE4CzgK1VtaeqngS2AqvbttdW1e1VVcANQ31JkmbITD8TOb6qHm3L3wOOb8uLgEeG2u1stf3Vd05Rn1KSdUm2J9m+e/fugzsDSdK/mrUH6+0KomboWNdW1cqqWrlw4cKZOKQkzQszHSKPtVtRtL+Pt/ouYMlQu8Wttr/64inqkqQZNNMhsgmYfMNqLXDzUP3C9pbWKuDpdttrC3BmkmPbA/UzgS1t2zNJVrW3si4c6kuSNEOOHFfHSW4ETgeOS7KTwVtWvwvclORi4LvAu1rzzcA5wATwfeAigKrak+RjwLbW7oqqmnxY/5sM3gB7JfBX7SNJmkFjC5GqumCaTe+Yom0Bl0zTz3pg/RT17cBPHcwYJUkHx2+sS5K6GSKSpG6GiCSpmyEiSepmiEiSuhkikqRuhogkqZshIknqZohIkroZIpKkboaIJKmbISJJ6maISJK6GSKSpG6GiCSpmyEiSepmiEiSuhkikqRuhogkqZshIknqZohIkroZIpKkboaIJKmbISJJ6maISJK6zUqIJPlOknuT3J1ke6u9PsnWJA+1v8e2epJcnWQiyT1JThnqZ21r/1CStbNxLpI0n83mlch/qKoVVbWyrV8G3FJVy4Bb2jrA2cCy9lkHXAOD0AEuB04DTgUunwweSdLMOJxuZ60BNrTlDcC5Q/UbauB24JgkJwBnAVurak9VPQlsBVbP9KAlaT6brRAp4G+S3JVkXasdX1WPtuXvAce35UXAI0P77my16eovkWRdku1Jtu/evftQnYMkzXtHztJxf66qdiX5MWBrkv89vLGqKkkdqoNV1bXAtQArV648ZP1K0nw3K1ciVbWr/X0c+DKDZxqPtdtUtL+Pt+a7gCVDuy9utenqkqQZMuMhkuRVSV4zuQycCdwHbAIm37BaC9zcljcBF7a3tFYBT7fbXluAM5Mc2x6on9lqkqQZMhu3s44Hvpxk8vj/o6r+Osk24KYkFwPfBd7V2m8GzgEmgO8DFwFU1Z4kHwO2tXZXVNWemTsNSdKMh0hVPQy8dYr6E8A7pqgXcMk0fa0H1h/qMUqSRnM4veIrSZpjDBFJUjdDRJLUzRCRJHUzRCRJ3QwRSVI3Q0SS1M0QkSR1M0QkSd0MEUlSN0NEktTNEJEkdTNEJEndDBFJUjdDRJLUzRCRJHUzRCRJ3QwRSVI3Q0SS1M0QkSR1M0QkSd0MEUlSN0NEktTNEJEkdTNEJEndDBFJUrc5HyJJVid5MMlEkstmezySNJ/M6RBJsgD4DHA2sBy4IMny2R2VJM0fczpEgFOBiap6uKp+AGwE1szymCRp3jhytgdwkBYBjwyt7wRO27dRknXAurb6XJIHZ2Bs88FxwD/O9iAOB/nk2tkegl7Kfz8nXZ5D0csbpyrO9RAZSVVdC1w72+N4uUmyvapWzvY4pKn47+fMmOu3s3YBS4bWF7eaJGkGzPUQ2QYsS3JSkqOA84FNszwmSZo35vTtrKram+RSYAuwAFhfVTtmeVjzibcIdTjz388ZkKqa7TFIkuaouX47S5I0iwwRSVI3Q0RdnG5Gh6sk65M8nuS+2R7LfGCI6IA53YwOc9cDq2d7EPOFIaIeTjejw1ZVfRXYM9vjmC8MEfWYarqZRbM0FkmzyBCRJHUzRNTD6WYkAYaI+jjdjCTAEFGHqtoLTE438wBwk9PN6HCR5EbgG8BPJtmZ5OLZHtPLmdOeSJK6eSUiSepmiEiSuhkikqRuhogkqZshIknqZohIkroZItIIknx9lo77zsmp9pOcOzxbcpIrkvzCbIxLmuT3RKQ5Isn1wFeq6ouzPRZpklci0giSPNf+npDkq0nuTnJfkn+/v32SXJVkR5Jbkixs9RVJbk9yT5IvJzm21X8ryf2tvrHV3pvkD5P8LPBO4Pfasd+U5Pok57UfCPvC0HFPT/KVtnxmkm8k+WaSLyR59fj+KWk+MkSkA/NuYEtVrQDeCty9n7avArZX1VuA24DLW/0G4ENV9TPAvUP1y4CTW/0/D3dUVV9nMD/ZB6tqRVX9w9DmvwVOS/Kqtv4fgY1JjgM+AvxCVZ0CbAf+a89JS9MxRKQDsw24KMlHgZ+uqmf30/ZfgM+35T8Hfi7J64Bjquq2Vt8A/Hxbvgf4XJJfB/aOOqA2l9lfA7+c5Ejgl4CbgVUMfnnyfyW5G1gLvHHUfqVRGCLSAWi/mvfzDKa+vz7JhQey+w/Z/ksMfnb4FGBbC4RRbQTeBZzB4OrnWSDA1nblsqKqlleVkxHqkDJEpAOQ5I3AY1X1WeBPGPwHfzpHAOe15XcDX6uqp4Enh56lvAe4LckRwJKquhX4EPA6YN/nF88Cr5nmWLe1sbyPQaAA3A68Pcmb29hfleQnRjtTaTQH8n86kuB04INJ/h/wHLC/K5F/Ak5N8hHgcQbPKmBwW+mPk/wo8DBwEbAA+PN2uyvA1VX1VJLh/jYCn03yW7wYTgBU1QvtYfp7W/9U1e4k7wVuTHJ0a/oR4P90nLc0JV/xlcYkyXNV5dtQelnzdpYkqZu3s6SDlOQO4Oh9yu/xKkTzgbezJEndvJ0lSepmiEiSuhkikqRuhogkqdv/B7v+kIzQgpWYAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "sgvTNVIGO40Q"
      },
      "source": [
        "from sklearn.model_selection import train_test_split"
      ],
      "execution_count": 80,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OyVsuwImO7a5"
      },
      "source": [
        "# splitting dataset using stratified sampling\n",
        "x_train, x_val, y_train, y_val = train_test_split(reviews.review, \n",
        "                                                  reviews.is_positive, \n",
        "                                                  test_size=0.30, \n",
        "                                                  random_state=17,\n",
        "                                                  stratify=reviews.is_positive.values\n",
        "                                               )"
      ],
      "execution_count": 81,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5aDvJDjrO9Xh"
      },
      "source": [
        "import pandas as pd\n",
        "data_train = pd.DataFrame()\n",
        "data_test  = pd.DataFrame()\n",
        "data_train['label'] = y_train\n",
        "data_train['text'] = x_train\n",
        "data_test['label'] = y_val\n",
        "data_test['text'] = x_val"
      ],
      "execution_count": 82,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QeH7IDosO_tm",
        "outputId": "fba81f1e-2254-498a-d9da-ee0e88dca070",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 143
        }
      },
      "source": [
        "train_sentences = data_train['text'].tolist()\n",
        "train_labels = data_train['label'].tolist()\n",
        "test_sentences = data_test['text'].tolist()\n",
        "test_labels = data_test['label'].tolist()\n",
        "data_train.groupby(['label']).count()"
      ],
      "execution_count": 83,
      "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>text</th>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>label</th>\n",
              "      <th></th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>21000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>21000</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "        text\n",
              "label       \n",
              "0      21000\n",
              "1      21000"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 83
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SM-3j5WbPCTI",
        "outputId": "ae6e1fee-ef9e-4a1d-94ba-596645ce443a",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "#Ratio of +ve and -ve reviews in training dataset\n",
        "ratio = str(data_train['label'].value_counts()[0]/data_train['label'].value_counts()[1])\n",
        "print(\"ratio of +ve to -ve dataEntry- \"+ratio+\":1\")"
      ],
      "execution_count": 84,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "ratio of +ve to -ve dataEntry- 1.0:1\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qyV61wAvPEmX",
        "outputId": "85ebbb90-4ff3-4084-a58a-e5c7cbe7942c",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "y_train.head(n=10)"
      ],
      "execution_count": 85,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "15393    0\n",
              "12330    1\n",
              "18976    0\n",
              "26737    1\n",
              "31241    1\n",
              "33259    0\n",
              "10836    0\n",
              "30247    0\n",
              "4817     0\n",
              "15889    0\n",
              "Name: is_positive, dtype: int64"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 85
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mrfGpRhZP4Ie",
        "outputId": "805042ca-3147-4dbb-eb3e-75690db69a91",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "!pip install bert-for-tf2"
      ],
      "execution_count": 86,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Requirement already satisfied: bert-for-tf2 in /usr/local/lib/python3.6/dist-packages (0.14.7)\n",
            "Requirement already satisfied: params-flow>=0.8.0 in /usr/local/lib/python3.6/dist-packages (from bert-for-tf2) (0.8.2)\n",
            "Requirement already satisfied: py-params>=0.9.6 in /usr/local/lib/python3.6/dist-packages (from bert-for-tf2) (0.9.7)\n",
            "Requirement already satisfied: numpy in /usr/local/lib/python3.6/dist-packages (from params-flow>=0.8.0->bert-for-tf2) (1.18.5)\n",
            "Requirement already satisfied: tqdm in /usr/local/lib/python3.6/dist-packages (from params-flow>=0.8.0->bert-for-tf2) (4.41.1)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YcTd_a8UQCzf",
        "outputId": "cefb1345-3560-4edf-a8e6-4fa186259f9a",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "!pip install sentencepiece"
      ],
      "execution_count": 87,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Requirement already satisfied: sentencepiece in /usr/local/lib/python3.6/dist-packages (0.1.94)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OudW_qHPQF_5",
        "outputId": "e380feb8-4732-49d4-b2a1-f828ff4ed343",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "import bert\n",
        "!pip install \"tensorflow_hub>=0.6.0\"\n",
        "import tensorflow_hub as hub"
      ],
      "execution_count": 88,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Requirement already satisfied: tensorflow_hub>=0.6.0 in /usr/local/lib/python3.6/dist-packages (0.10.0)\n",
            "Requirement already satisfied: numpy>=1.12.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow_hub>=0.6.0) (1.18.5)\n",
            "Requirement already satisfied: protobuf>=3.8.0 in /usr/local/lib/python3.6/dist-packages (from tensorflow_hub>=0.6.0) (3.12.4)\n",
            "Requirement already satisfied: setuptools in /usr/local/lib/python3.6/dist-packages (from protobuf>=3.8.0->tensorflow_hub>=0.6.0) (50.3.2)\n",
            "Requirement already satisfied: six>=1.9 in /usr/local/lib/python3.6/dist-packages (from protobuf>=3.8.0->tensorflow_hub>=0.6.0) (1.15.0)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cmekLEUaQQC3"
      },
      "source": [
        "BertTokenizer = bert.bert_tokenization.FullTokenizer"
      ],
      "execution_count": 89,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SnPn8z8iQR81"
      },
      "source": [
        "model_url = \"https://tfhub.dev/tensorflow/bert_en_cased_L-12_H-768_A-12/1\"\n",
        "bert_layer = hub.KerasLayer(model_url, trainable=False)\n",
        "vocab_file = bert_layer.resolved_object.vocab_file.asset_path.numpy()"
      ],
      "execution_count": 90,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wAiOMJ2-QUrA"
      },
      "source": [
        "to_lower_case = bert_layer.resolved_object.do_lower_case.numpy()\n",
        "tokenizer = BertTokenizer(vocab_file, to_lower_case)"
      ],
      "execution_count": 91,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "A9VX24tBQZ_D",
        "outputId": "9d75e1ea-fa47-49fe-e838-0fb359be17c7",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "tokenizer.tokenize(\"This hotel is sooo nice\")\n",
        "print(len(tokenizer.vocab))"
      ],
      "execution_count": 92,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "28996\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Q7oi5CKjQcWB"
      },
      "source": [
        "def tokenize_reviews(text_review):\n",
        "    return tokenizer.convert_tokens_to_ids(tokenizer.tokenize(text_review))"
      ],
      "execution_count": 93,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LLI3PqqOQfhC"
      },
      "source": [
        "tokenized_review = [tokenize_reviews(review) for review in reviews_text]"
      ],
      "execution_count": 94,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "g4-JRarfQiqF"
      },
      "source": [
        "vocab_size = len(tokenizer.vocab) + 1\n",
        "embedding_dim = 128\n",
        "max_length = 128\n",
        "trunc_type='post'\n",
        "padding_type='post'\n",
        "oov_tok = \"<OOV>\"\n",
        "training_portion=0.7"
      ],
      "execution_count": 95,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rR57aCjUQ-3r"
      },
      "source": [
        "train_tokens = list(map(lambda t: ['[CLS]'] + tokenizer.tokenize(t)[:127], train_sentences))\n",
        "test_tokens = list(map(lambda t: ['[CLS]'] + tokenizer.tokenize(t)[:127], test_sentences))"
      ],
      "execution_count": 96,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "iER6V61WRN4k"
      },
      "source": [
        "train_tokens_ids = list(map(tokenizer.convert_tokens_to_ids, train_tokens))"
      ],
      "execution_count": 97,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Ef7R9_PzTYV0"
      },
      "source": [
        "test_tokens_ids = list(map(tokenizer.convert_tokens_to_ids, test_tokens))"
      ],
      "execution_count": 98,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wD4YwOoATfIb"
      },
      "source": [
        "train_padded = tf.keras.preprocessing.sequence.pad_sequences(train_tokens_ids, maxlen=128, truncating=\"post\", padding=\"post\", dtype=\"int\")\n",
        "test_padded = tf.keras.preprocessing.sequence.pad_sequences(test_tokens_ids, maxlen=128, truncating=\"post\", padding=\"post\", dtype=\"int\")"
      ],
      "execution_count": 99,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DIpddVAkT_VE"
      },
      "source": [
        "train_padded = np.array(train_padded)\n",
        "train_labels = np.array(train_labels)\n",
        "test_padded = np.array(test_padded)\n",
        "test_labels = np.array(test_labels)"
      ],
      "execution_count": 100,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TGCMvt7DEy3H"
      },
      "source": [
        "from keras.models import Model\n",
        "from keras.layers import Input\n",
        "from keras.layers import Dense\n",
        "from keras.layers import Flatten\n",
        "from keras.layers import Dropout\n",
        "from keras.layers import Embedding\n",
        "from keras.layers.convolutional import Conv1D\n",
        "from keras.layers.convolutional import MaxPooling1D\n",
        "from keras.layers.merge import concatenate\n",
        "from keras.utils.vis_utils import plot_model"
      ],
      "execution_count": 101,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "824BuvegeurW",
        "outputId": "e05c8f04-748b-4b4f-b3a7-ae99d4dd7af4",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "print(len(train_labels))"
      ],
      "execution_count": 102,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "42000\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "h8YAFoLwZ2sW",
        "outputId": "4acea8c6-f84b-49ca-d8fa-0a2682449b66",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "model = tf.keras.Sequential([\n",
        "  tf.keras.layers.Embedding(vocab_size, embedding_dim, input_length=max_length),\n",
        "  tf.keras.layers.Conv1D(128, 5, activation='relu'),\n",
        "  tf.keras.layers.Bidirectional(tf.keras.layers.LSTM(32)),\n",
        "  tf.keras.layers.Dense(24, activation='relu'),\n",
        "  tf.keras.layers.Dense(1, activation='sigmoid' )\n",
        "])\n",
        "\n",
        "#compile model\n",
        "model.compile(loss=\"binary_crossentropy\",\n",
        "             optimizer=\"adam\",\n",
        "             metrics=['accuracy'])\n",
        "#model.build(input_shape=[len(train_padded),512,128])\n",
        "\n",
        "model.summary()"
      ],
      "execution_count": 103,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential_1\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "embedding_1 (Embedding)      (None, 128, 128)          3711616   \n",
            "_________________________________________________________________\n",
            "conv1d_1 (Conv1D)            (None, 124, 128)          82048     \n",
            "_________________________________________________________________\n",
            "bidirectional_1 (Bidirection (None, 64)                41216     \n",
            "_________________________________________________________________\n",
            "dense_2 (Dense)              (None, 24)                1560      \n",
            "_________________________________________________________________\n",
            "dense_3 (Dense)              (None, 1)                 25        \n",
            "=================================================================\n",
            "Total params: 3,836,465\n",
            "Trainable params: 3,836,465\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "uSzBIn-iUvSt"
      },
      "source": [
        "length = 128\n",
        "vocab_size = len(tokenizer.vocab) + 1"
      ],
      "execution_count": 104,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "oSrySzRBU5-S",
        "outputId": "17cd41f8-a287-4b76-a52e-821530d969f8",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "history = model.fit(train_padded, train_labels, verbose=1, epochs=10, validation_data=(test_padded, test_labels))"
      ],
      "execution_count": 105,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/10\n",
            "1313/1313 [==============================] - 212s 161ms/step - loss: 0.2172 - accuracy: 0.9171 - val_loss: 0.2005 - val_accuracy: 0.9260\n",
            "Epoch 2/10\n",
            "1313/1313 [==============================] - 214s 163ms/step - loss: 0.1559 - accuracy: 0.9440 - val_loss: 0.2018 - val_accuracy: 0.9261\n",
            "Epoch 3/10\n",
            "1313/1313 [==============================] - 208s 158ms/step - loss: 0.1260 - accuracy: 0.9564 - val_loss: 0.2165 - val_accuracy: 0.9253\n",
            "Epoch 4/10\n",
            "1313/1313 [==============================] - 213s 162ms/step - loss: 0.1039 - accuracy: 0.9644 - val_loss: 0.2292 - val_accuracy: 0.9232\n",
            "Epoch 5/10\n",
            "1313/1313 [==============================] - 212s 161ms/step - loss: 0.0864 - accuracy: 0.9704 - val_loss: 0.2526 - val_accuracy: 0.9254\n",
            "Epoch 6/10\n",
            "1313/1313 [==============================] - 208s 159ms/step - loss: 0.0762 - accuracy: 0.9736 - val_loss: 0.2821 - val_accuracy: 0.9241\n",
            "Epoch 7/10\n",
            "1313/1313 [==============================] - 210s 160ms/step - loss: 0.0681 - accuracy: 0.9766 - val_loss: 0.3052 - val_accuracy: 0.9241\n",
            "Epoch 8/10\n",
            "1313/1313 [==============================] - 223s 170ms/step - loss: 0.0615 - accuracy: 0.9788 - val_loss: 0.3187 - val_accuracy: 0.9228\n",
            "Epoch 9/10\n",
            "1313/1313 [==============================] - 221s 169ms/step - loss: 0.0567 - accuracy: 0.9802 - val_loss: 0.3437 - val_accuracy: 0.9198\n",
            "Epoch 10/10\n",
            "1313/1313 [==============================] - 216s 165ms/step - loss: 0.0522 - accuracy: 0.9813 - val_loss: 0.3652 - val_accuracy: 0.9217\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BHuKP4XEIvlH"
      },
      "source": [
        "from sklearn.metrics import confusion_matrix\n",
        "x_pred = model.predict(test_padded)\n",
        "x_pred = x_pred > 0.5\n",
        "x_pred = x_pred.astype(int)\n",
        "y_pred = test_labels"
      ],
      "execution_count": 108,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-Dw8tDT-3qXc",
        "outputId": "ebbc1201-1324-461f-8cd8-688ba5638988",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "from sklearn.metrics import f1_score, precision_score, recall_score, confusion_matrix,accuracy_score\n",
        "print(\"F1-score: {0}\".format(f1_score(x_pred, y_pred)))\n",
        "print(\"Precision: \",precision_score(y_pred, y_pred))\n",
        "print(\"Recall: \",recall_score(y_pred, y_pred))\n",
        "print(\"Accuracy: \",accuracy_score(y_pred,y_pred))\n",
        "print(\"Confusion matrix: \")\n",
        "confusion_matrix(y_pred, y_pred)"
      ],
      "execution_count": 109,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "F1-score: 0.9215358931552587\n",
            "Precision:  1.0\n",
            "Recall:  1.0\n",
            "Accuracy:  1.0\n",
            "Confusion matrix: \n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([[9000,    0],\n",
              "       [   0, 9000]])"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 109
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "piZZ6yBBVX75"
      },
      "source": [
        ""
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "w63UV7CFU8QO",
        "outputId": "d038d2a6-5171-44f2-9142-ec650c2e51a8",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 651
        }
      },
      "source": [
        "plt.figure(figsize=(10,5))\n",
        "plt.subplot(1,2,1)\n",
        "plt.plot(history.history['accuracy'], label='Training Accuracy')\n",
        "plt.plot(history.history['val_accuracy'], label='Validation Accuracy')\n",
        "plt.xlabel(\"Epochs\")\n",
        "plt.ylabel(\"Accuracy\")\n",
        "\n",
        "plt.figure(figsize=(10,5))\n",
        "plt.subplot(1,2,1)\n",
        "plt.plot(history.history['loss'], label='Training Loss')\n",
        "plt.plot(history.history['val_loss'], label='Validation Loss')\n",
        "plt.xlabel(\"Epochs\")\n",
        "plt.ylabel(\"Loss\")\n",
        "\n",
        "plt.legend()\n",
        "plt.show()"
      ],
      "execution_count": 107,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x360 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x360 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    }
  ]
}