sentiment_analysis_notebook.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "sentiment_analysis_notebook.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "mount_file_id": "1xcn_mQxLK1tX6Pg0zVLtEPV_ua0ahtDY",
      "authorship_tag": "ABX9TyOGMSponp1EqNJSSkv8acku",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/ArunRK7Codie/Sentiment-Analysis/blob/master/sentiment_analysis_notebook.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ej1Vmrdbh1Cr",
        "colab_type": "code",
        "outputId": "cdc93ed7-2561-4528-a44e-0456a601ff66",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 71
        }
      },
      "source": [
        "import tensorflow as tf\n",
        "import matplotlib.pyplot as plt\n",
        "import pandas as pd\n",
        "import numpy as np\n",
        "\n",
        "\n",
        "import nltk \n",
        "nltk.download('stopwords')\n",
        "from nltk.corpus import stopwords\n",
        "from nltk.stem import SnowballStemmer\n",
        "\n",
        "\n",
        "from keras.preprocessing.text import Tokenizer\n",
        "from keras.preprocessing.sequence import pad_sequences\n",
        "\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.preprocessing import LabelEncoder\n",
        "\n",
        "\n",
        "import re"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[nltk_data] Downloading package stopwords to /root/nltk_data...\n",
            "[nltk_data]   Unzipping corpora/stopwords.zip.\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "Using TensorFlow backend.\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rqL3x4B5jsOw",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import requests\n",
        "r = requests.get(\"http://cs.stanford.edu/people/alecmgo/trainingandtestdata.zip\")\n",
        "with open('dataset.zip','wb') as f:\n",
        "  f.write(r.content)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "u7A40z28kFdW",
        "colab_type": "code",
        "outputId": "438173f5-c218-436e-efe7-f9f51bf37bcd",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        }
      },
      "source": [
        "from zipfile import ZipFile\n",
        "with ZipFile('dataset.zip', 'r') as zip:\n",
        "  zip.extractall()\n",
        "  print('Files Extracted')"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Files Extracted\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "B0adTKHGGj7a",
        "colab_type": "code",
        "outputId": "f6fe31ee-0304-4ce3-9733-508b0b522ad2",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 204
        }
      },
      "source": [
        "df = pd.read_csv('/content/training.1600000.processed.noemoticon.csv',\n",
        "                 encoding = 'latin',header=None)\n",
        "df.head()"
      ],
      "execution_count": 0,
      "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>0</th>\n",
              "      <th>1</th>\n",
              "      <th>2</th>\n",
              "      <th>3</th>\n",
              "      <th>4</th>\n",
              "      <th>5</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0</td>\n",
              "      <td>1467810369</td>\n",
              "      <td>Mon Apr 06 22:19:45 PDT 2009</td>\n",
              "      <td>NO_QUERY</td>\n",
              "      <td>_TheSpecialOne_</td>\n",
              "      <td>@switchfoot http://twitpic.com/2y1zl - Awww, t...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>0</td>\n",
              "      <td>1467810672</td>\n",
              "      <td>Mon Apr 06 22:19:49 PDT 2009</td>\n",
              "      <td>NO_QUERY</td>\n",
              "      <td>scotthamilton</td>\n",
              "      <td>is upset that he can't update his Facebook by ...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>0</td>\n",
              "      <td>1467810917</td>\n",
              "      <td>Mon Apr 06 22:19:53 PDT 2009</td>\n",
              "      <td>NO_QUERY</td>\n",
              "      <td>mattycus</td>\n",
              "      <td>@Kenichan I dived many times for the ball. Man...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>0</td>\n",
              "      <td>1467811184</td>\n",
              "      <td>Mon Apr 06 22:19:57 PDT 2009</td>\n",
              "      <td>NO_QUERY</td>\n",
              "      <td>ElleCTF</td>\n",
              "      <td>my whole body feels itchy and like its on fire</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>0</td>\n",
              "      <td>1467811193</td>\n",
              "      <td>Mon Apr 06 22:19:57 PDT 2009</td>\n",
              "      <td>NO_QUERY</td>\n",
              "      <td>Karoli</td>\n",
              "      <td>@nationwideclass no, it's not behaving at all....</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   0  ...                                                  5\n",
              "0  0  ...  @switchfoot http://twitpic.com/2y1zl - Awww, t...\n",
              "1  0  ...  is upset that he can't update his Facebook by ...\n",
              "2  0  ...  @Kenichan I dived many times for the ball. Man...\n",
              "3  0  ...    my whole body feels itchy and like its on fire \n",
              "4  0  ...  @nationwideclass no, it's not behaving at all....\n",
              "\n",
              "[5 rows x 6 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 4
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "l5po3MbiL0mb",
        "colab_type": "code",
        "outputId": "9e3101a7-d89a-4121-ad0e-88c3e72cfeb0",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 204
        }
      },
      "source": [
        "df.columns = ['sentiment', 'id', 'date', 'query', 'user_id', 'text']\n",
        "df.head()"
      ],
      "execution_count": 0,
      "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>sentiment</th>\n",
              "      <th>id</th>\n",
              "      <th>date</th>\n",
              "      <th>query</th>\n",
              "      <th>user_id</th>\n",
              "      <th>text</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0</td>\n",
              "      <td>1467810369</td>\n",
              "      <td>Mon Apr 06 22:19:45 PDT 2009</td>\n",
              "      <td>NO_QUERY</td>\n",
              "      <td>_TheSpecialOne_</td>\n",
              "      <td>@switchfoot http://twitpic.com/2y1zl - Awww, t...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>0</td>\n",
              "      <td>1467810672</td>\n",
              "      <td>Mon Apr 06 22:19:49 PDT 2009</td>\n",
              "      <td>NO_QUERY</td>\n",
              "      <td>scotthamilton</td>\n",
              "      <td>is upset that he can't update his Facebook by ...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>0</td>\n",
              "      <td>1467810917</td>\n",
              "      <td>Mon Apr 06 22:19:53 PDT 2009</td>\n",
              "      <td>NO_QUERY</td>\n",
              "      <td>mattycus</td>\n",
              "      <td>@Kenichan I dived many times for the ball. Man...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>0</td>\n",
              "      <td>1467811184</td>\n",
              "      <td>Mon Apr 06 22:19:57 PDT 2009</td>\n",
              "      <td>NO_QUERY</td>\n",
              "      <td>ElleCTF</td>\n",
              "      <td>my whole body feels itchy and like its on fire</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>0</td>\n",
              "      <td>1467811193</td>\n",
              "      <td>Mon Apr 06 22:19:57 PDT 2009</td>\n",
              "      <td>NO_QUERY</td>\n",
              "      <td>Karoli</td>\n",
              "      <td>@nationwideclass no, it's not behaving at all....</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   sentiment  ...                                               text\n",
              "0          0  ...  @switchfoot http://twitpic.com/2y1zl - Awww, t...\n",
              "1          0  ...  is upset that he can't update his Facebook by ...\n",
              "2          0  ...  @Kenichan I dived many times for the ball. Man...\n",
              "3          0  ...    my whole body feels itchy and like its on fire \n",
              "4          0  ...  @nationwideclass no, it's not behaving at all....\n",
              "\n",
              "[5 rows x 6 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 5
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lssiD2RQMsxc",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "df = df.drop(['id', 'date', 'query', 'user_id'], axis=1)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cL17bzZiPFtN",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "lab_to_sentiment = {0:\"Negative\", 4:\"Positive\"}\n",
        "def label_decoder(label):\n",
        "  return lab_to_sentiment[label]\n",
        "df.sentiment = df.sentiment.apply(lambda x: label_decoder(x))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AV4wS9uXGqlP",
        "colab_type": "code",
        "outputId": "3e5c37a0-cbd8-423e-fbdd-47aed9ba2cb3",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 204
        }
      },
      "source": [
        "df.head()"
      ],
      "execution_count": 0,
      "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>sentiment</th>\n",
              "      <th>text</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>Negative</td>\n",
              "      <td>@switchfoot http://twitpic.com/2y1zl - Awww, t...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>Negative</td>\n",
              "      <td>is upset that he can't update his Facebook by ...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>Negative</td>\n",
              "      <td>@Kenichan I dived many times for the ball. Man...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>Negative</td>\n",
              "      <td>my whole body feels itchy and like its on fire</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>Negative</td>\n",
              "      <td>@nationwideclass no, it's not behaving at all....</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "  sentiment                                               text\n",
              "0  Negative  @switchfoot http://twitpic.com/2y1zl - Awww, t...\n",
              "1  Negative  is upset that he can't update his Facebook by ...\n",
              "2  Negative  @Kenichan I dived many times for the ball. Man...\n",
              "3  Negative    my whole body feels itchy and like its on fire \n",
              "4  Negative  @nationwideclass no, it's not behaving at all...."
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 8
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6uAGLkyQLzNh",
        "colab_type": "code",
        "outputId": "ca58cfc7-a714-4b2d-9e65-b21d180f80db",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 299
        }
      },
      "source": [
        "val_count = df.sentiment.value_counts()\n",
        "\n",
        "plt.figure(figsize=(8,4))\n",
        "plt.bar(val_count.index, val_count.values)\n",
        "plt.title(\"Sentiment Data Distribution\")"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Text(0.5, 1.0, 'Sentiment Data Distribution')"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 9
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfkAAAEICAYAAABceI1YAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+j8jraAAAe8ElEQVR4nO3dfbxdVX3n8c9XIoIPPGdQEzS0plpkhEqEWLU6YiGgbWhHEYeWSCnRitrWdiraVhSlxZmpClPFYYQSWiug1ZIqElPAqdqCREUQqCVFMYk8xIQHkWf6mz/2unq4nnvPSUhI2Hzer9d53b1/a+211rk3N9+z99n33lQVkiSpfx63pRcgSZI2D0NekqSeMuQlSeopQ16SpJ4y5CVJ6ilDXpKknjLkpUdAko8m+dMtvY5HsySfT7JoE431kiTfHtj/bpJXbIqx23hXJ3nZphpP2liGvB6zkrw4yT8nuT3J+iRfSfKCTTDu65N8ebBWVW+sqvc+3LE3Yi3vTvI3I/p8N8ndSX6Y5Lb2OXljkrH+f0gyJ0klmfEw1llJfpTkziTrklyU5LWDfarqkKpaMuZYz5quT1V9qaqevbHrnTTfWUneN2n851bVFzfF+NLDYcjrMSnJDsBngf8N7ALMAt4D3Lsl17UF/UpVPQV4JnAy8HbgjEd4DftU1ZOBZwNnAX+Z5IRNPcnDeTEiPepUlQ8fj7kHMA+4bUSf3wKuBW4FlgHPHGgr4I3AdcBtwIeBAD8P3AM8CNw5MQddaL2vbb8MWA38EXALcCNwGHAo8G/AeuCdA3M9Djge+HdgHXAesEtrm9PWsgj4HvAD4I9b2wLgPuD+tpZvTvE8vwu8YlJtf+A/gL3b/iuBbwB3AKuAdw/0/V5bw53t8ULgZ4GL23p/AHwc2Gmaz3UBz5pUe3X7XO7a9r8I/Hbbfhbw/4Db2/jntvo/tbF+1Nby2oHP99uBm4C/nqhN+hy8A7imfb3/Ctiutb0e+PKw9QKL2+f3vjbfP0z+nAJPAD4EfL89PgQ8YdK/hT8Y+Ldw9Jb+/vDRn4dn8nqs+jfgwSRLkhySZOfBxiQLgXcCvw7MBL4EfGLSGK8CXgA8DzgcOLiqrqUL/3+pqidX1U5TzP9UYDu6KwjvAv4v8BvAfsBLgD9Nsmfr+xa6FwEvBZ5OF0IfnjTei+nOgA8E3pXk56vqQuDP6ALwyVW1z3ifGqiqr9KFz0ta6UfAUcBOdIH/O0kOa22/1D7u1Ob5F7oXPH/e1vvzwB7Au8edvzkfmEH3gmOy9wJfAHYGZtNdkaGqJtayT1vLuW3/qXRXbJ5JF8zDHAkcTPcC5eeAPxm1wKo6ne4FzP9o8/3KkG5/DMwH9gX2ac9ncOynAjvS/Vs4Bvjw5H+P0sYy5PWYVFV30AVj0QXs2iRLk+zeurwR+POquraqHqALy32TPHNgmJOr6raq+h5wCd1/4uO6Hzipqu4HzgF2A06pqh9W1dV0Z5QTofxGurPz1VV1L11YvnrSZef3VNXdVfVN4JsDxz4c36cLRqrqi1V1VVX9R1VdSfeC56VTHVhVK6tqeVXdW1VrgQ9M13+KMe6nO0vfZUjz/XSB/fSquqeqvjykz6D/AE5o67l7ij5/WVWrqmo9cBLwug1Z7zSOBE6sqlva5+I9wG8OtN/f2u+vqgvorghskvsFJENej1ktwF9fVbOBvenOOj/Ump8JnNJuRLuN7hJ66M62Jtw0sH0X8OQNmH5dVT3YtidC5+aB9rsHxnsm8JmBtVxL93bA7gP9H85apjKL7nmT5IAklyRZm+R2uhceu011YJLdk5yTZE2SO4C/ma7/FGM8nu4qyvohzX9E9/X4aruT/bdGDLe2qu4Z0WfVwPYNdP8eNoWnt/GmGntdeyE5YVN9/SRDXgKoqn+le99871ZaBbyhqnYaeGxfVf88znCbeHmrgEMmrWW7qlqzudbSfspgFjBxhvy3wFJgj6raEfgoXchONceftfp/rqod6N6KyJB+01kIPAB8dXJDVd1UVcdW1dOBNwAfGXFH/Tifhz0Gtp9BdyUDurcqnjjRkOSpGzj29+leqA0bW9qsDHk9JiV5TpI/SDK77e9Bd3n20tblo8A7kjy3te+Y5DVjDn8zMDvJtptouR8FTpp4qyDJzHbPwLhrmbMBPw63Q5JX0b2F8DdVdVVregqwvqruSbI/8N8GDltLdzn8ZwZqT6G77Hx7klnAfx9zvSTZJcmRdPcdvL+q1g3p85qJrx3dPQrV1gDdc/6ZyceM4bgks5PsQvc++sT7+d8Enptk3yTb8dP3Foya7xPAn7Sv225092BM+2ON0qZiyOux6ofAAcBlSX5EF+7forvLmar6DPB+4Jx2uflbwCFjjn0xcDVwU5IfbIK1nkJ3Fv2FJD9saz1gzGM/2T6uS/L1afr9Qxt7FV3AfQA4eqD9TcCJrc+76O7wB6Cq7qJ7D/sr7S2F+XTvOz+f7u73zwGfHmOt30xyJ7AS+G3g96vqXVP0fQHd1+5Ous/N71bV9a3t3cCStpbDx5h3wt/S3cx3Pd1PMryvPb9/A04E/pHupykmv/9/BrBXm+/vh4z7PmAFcCVwFfD1ibGlzS1Vm/rKoiRJ2hp4Ji9JUk8Z8pIk9ZQhL0lSTxnykiT1VO/+UMNuu+1Wc+bM2dLLkCTpEfG1r33tB1U1c1hb70J+zpw5rFixYksvQ5KkR0SSG6Zq83K9JEk9ZchLktRThrwkST1lyEuS1FOGvCRJPWXIS5LUU+P++cnfT3J1km8l+USS7ZLsmeSyJCuTnDvxZzWTPKHtr2ztcwbGeUerfzvJwQP1Ba22MsnxA/Whc0iSpNFGhnz7W9BvBeZV1d7ANsARdH+G84NV9Sy6v+d8TDvkGODWVv9g60eSvdpxzwUWAB9Jsk2Sbej+bvQhwF7A61pfpplDkiSNMO7l+hnA9klmAE8EbgReDnyqtS8BDmvbC9s+rf3AJGn1c6rq3qr6Dt3fjN6/PVZW1fVVdR9wDrCwHTPVHJIkaYSRv/GuqtYk+V/A94C7gS8AXwNuq6oHWrfVwKy2PQtY1Y59IMntwK6tfunA0IPHrJpUP6AdM9UcD5FkMbAY4BnPeMaop7RB5hz/uU06nrQlfffkV27pJWwUvw/VJ4/k9+E4l+t3pjsL3xN4OvAkusvtW42qOr2q5lXVvJkzh/76XkmSHnPGuVz/CuA7VbW2qu4HPg28CNipXb4HmA2sadtrgD0AWvuOwLrB+qRjpqqvm2YOSZI0wjgh/z1gfpIntvfJDwSuAS4BXt36LALOb9tL2z6t/eKqqlY/ot19vycwF/gqcDkwt91Jvy3dzXlL2zFTzSFJkkYYGfJVdRndzW9fB65qx5wOvB14W5KVdO+fn9EOOQPYtdXfBhzfxrkaOI/uBcKFwHFV9WB7z/3NwDLgWuC81pdp5pAkSSOM9admq+oE4IRJ5evp7oyf3Pce4DVTjHMScNKQ+gXABUPqQ+eQJEmj+RvvJEnqKUNekqSeMuQlSeopQ16SpJ4y5CVJ6ilDXpKknjLkJUnqKUNekqSeMuQlSeopQ16SpJ4y5CVJ6ilDXpKknjLkJUnqKUNekqSeMuQlSeopQ16SpJ4aGfJJnp3kioHHHUl+L8kuSZYnua593Ln1T5JTk6xMcmWS5w+Mtaj1vy7JooH6fkmuasecmiStPnQOSZI02siQr6pvV9W+VbUvsB9wF/AZ4HjgoqqaC1zU9gEOAea2x2LgNOgCGzgBOADYHzhhILRPA44dOG5Bq081hyRJGmFDL9cfCPx7Vd0ALASWtPoS4LC2vRA4uzqXAjsleRpwMLC8qtZX1a3AcmBBa9uhqi6tqgLOnjTWsDkkSdIIGxryRwCfaNu7V9WNbfsmYPe2PQtYNXDM6labrr56SH26OSRJ0ghjh3ySbYFfBT45ua2dgdcmXNdPmW6OJIuTrEiyYu3atZtzGZIkPWpsyJn8IcDXq+rmtn9zu9RO+3hLq68B9hg4bnarTVefPaQ+3RwPUVWnV9W8qpo3c+bMDXhKkiT114aE/Ov4yaV6gKXAxB3yi4DzB+pHtbvs5wO3t0vuy4CDkuzcbrg7CFjW2u5IMr/dVX/UpLGGzSFJkkaYMU6nJE8Cfhl4w0D5ZOC8JMcANwCHt/oFwKHASro78Y8GqKr1Sd4LXN76nVhV69v2m4CzgO2Bz7fHdHNIkqQRxgr5qvoRsOuk2jq6u+0n9y3guCnGORM4c0h9BbD3kPrQOSRJ0mj+xjtJknrKkJckqacMeUmSesqQlySppwx5SZJ6ypCXJKmnDHlJknrKkJckqacMeUmSesqQlySppwx5SZJ6ypCXJKmnDHlJknrKkJckqacMeUmSesqQlySppwx5SZJ6aqyQT7JTkk8l+dck1yZ5YZJdkixPcl37uHPrmySnJlmZ5Mokzx8YZ1Hrf12SRQP1/ZJc1Y45NUlafegckiRptHHP5E8BLqyq5wD7ANcCxwMXVdVc4KK2D3AIMLc9FgOnQRfYwAnAAcD+wAkDoX0acOzAcQtafao5JEnSCCNDPsmOwC8BZwBU1X1VdRuwEFjSui0BDmvbC4Gzq3MpsFOSpwEHA8uran1V3QosBxa0th2q6tKqKuDsSWMNm0OSJI0wzpn8nsBa4K+SfCPJx5I8Cdi9qm5sfW4Cdm/bs4BVA8evbrXp6quH1JlmjodIsjjJiiQr1q5dO8ZTkiSp/8YJ+RnA84HTquoXgB8x6bJ5OwOvTb+88eaoqtOral5VzZs5c+bmXIYkSY8a44T8amB1VV3W9j9FF/o3t0vttI+3tPY1wB4Dx89utenqs4fUmWYOSZI0wsiQr6qbgFVJnt1KBwLXAEuBiTvkFwHnt+2lwFHtLvv5wO3tkvsy4KAkO7cb7g4ClrW2O5LMb3fVHzVprGFzSJKkEWaM2e8twMeTbAtcDxxN9wLhvCTHADcAh7e+FwCHAiuBu1pfqmp9kvcCl7d+J1bV+rb9JuAsYHvg8+0BcPIUc0iSpBHGCvmqugKYN6TpwCF9CzhuinHOBM4cUl8B7D2kvm7YHJIkaTR/450kST1lyEuS1FOGvCRJPWXIS5LUU4a8JEk9ZchLktRThrwkST1lyEuS1FOGvCRJPWXIS5LUU4a8JEk9ZchLktRThrwkST1lyEuS1FOGvCRJPWXIS5LUU4a8JEk9NVbIJ/lukquSXJFkRavtkmR5kuvax51bPUlOTbIyyZVJnj8wzqLW/7okiwbq+7XxV7ZjM90ckiRptA05k/8vVbVvVc1r+8cDF1XVXOCitg9wCDC3PRYDp0EX2MAJwAHA/sAJA6F9GnDswHELRswhSZJGeDiX6xcCS9r2EuCwgfrZ1bkU2CnJ04CDgeVVtb6qbgWWAwta2w5VdWlVFXD2pLGGzSFJkkYYN+QL+EKSryVZ3Gq7V9WNbfsmYPe2PQtYNXDs6labrr56SH26OR4iyeIkK5KsWLt27ZhPSZKkfpsxZr8XV9WaJP8JWJ7kXwcbq6qS1KZf3nhzVNXpwOkA8+bN26zrkCTp0WKsM/mqWtM+3gJ8hu499ZvbpXbax1ta9zXAHgOHz2616eqzh9SZZg5JkjTCyJBP8qQkT5nYBg4CvgUsBSbukF8EnN+2lwJHtbvs5wO3t0vuy4CDkuzcbrg7CFjW2u5IMr/dVX/UpLGGzSFJkkYY53L97sBn2k+1zQD+tqouTHI5cF6SY4AbgMNb/wuAQ4GVwF3A0QBVtT7Je4HLW78Tq2p9234TcBawPfD59gA4eYo5JEnSCCNDvqquB/YZUl8HHDikXsBxU4x1JnDmkPoKYO9x55AkSaP5G+8kSeopQ16SpJ4y5CVJ6ilDXpKknjLkJUnqKUNekqSeMuQlSeopQ16SpJ4y5CVJ6ilDXpKknjLkJUnqKUNekqSeMuQlSeopQ16SpJ4y5CVJ6ilDXpKknjLkJUnqqbFDPsk2Sb6R5LNtf88klyVZmeTcJNu2+hPa/srWPmdgjHe0+reTHDxQX9BqK5McP1AfOockSRptQ87kfxe4dmD//cAHq+pZwK3AMa1+DHBrq3+w9SPJXsARwHOBBcBH2guHbYAPA4cAewGva32nm0OSJI0wVsgnmQ28EvhY2w/wcuBTrcsS4LC2vbDt09oPbP0XAudU1b1V9R1gJbB/e6ysquur6j7gHGDhiDkkSdII457Jfwj4I+A/2v6uwG1V9UDbXw3MatuzgFUArf321v/H9UnHTFWfbo6HSLI4yYokK9auXTvmU5Ikqd9GhnySVwG3VNXXHoH1bJSqOr2q5lXVvJkzZ27p5UiStFWYMUafFwG/muRQYDtgB+AUYKckM9qZ9mxgTeu/BtgDWJ1kBrAjsG6gPmHwmGH1ddPMIUmSRhh5Jl9V76iq2VU1h+7GuYur6kjgEuDVrdsi4Py2vbTt09ovrqpq9SPa3fd7AnOBrwKXA3PbnfTbtjmWtmOmmkOSJI3wcH5O/u3A25KspHv//IxWPwPYtdXfBhwPUFVXA+cB1wAXAsdV1YPtLP3NwDK6u/fPa32nm0OSJI0wzuX6H6uqLwJfbNvX090ZP7nPPcBrpjj+JOCkIfULgAuG1IfOIUmSRvM33kmS1FOGvCRJPWXIS5LUU4a8JEk9ZchLktRThrwkST1lyEuS1FOGvCRJPWXIS5LUU4a8JEk9ZchLktRThrwkST1lyEuS1FOGvCRJPWXIS5LUU4a8JEk9ZchLktRTI0M+yXZJvprkm0muTvKeVt8zyWVJViY5N8m2rf6Etr+ytc8ZGOsdrf7tJAcP1Be02sokxw/Uh84hSZJGG+dM/l7g5VW1D7AvsCDJfOD9wAer6lnArcAxrf8xwK2t/sHWjyR7AUcAzwUWAB9Jsk2SbYAPA4cAewGva32ZZg5JkjTCyJCvzp1t9/HtUcDLgU+1+hLgsLa9sO3T2g9MklY/p6rurarvACuB/dtjZVVdX1X3AecAC9sxU80hSZJGGOs9+XbGfQVwC7Ac+Hfgtqp6oHVZDcxq27OAVQCt/XZg18H6pGOmqu86zRyT17c4yYokK9auXTvOU5IkqffGCvmqerCq9gVm0515P2ezrmoDVdXpVTWvqubNnDlzSy9HkqStwgbdXV9VtwGXAC8EdkoyozXNBta07TXAHgCtfUdg3WB90jFT1ddNM4ckSRphnLvrZybZqW1vD/wycC1d2L+6dVsEnN+2l7Z9WvvFVVWtfkS7+35PYC7wVeByYG67k35bupvzlrZjpppDkiSNMGN0F54GLGl3wT8OOK+qPpvkGuCcJO8DvgGc0fqfAfx1kpXAerrQpqquTnIecA3wAHBcVT0IkOTNwDJgG+DMqrq6jfX2KeaQJEkjjAz5qroS+IUh9evp3p+fXL8HeM0UY50EnDSkfgFwwbhzSJKk0fyNd5Ik9ZQhL0lSTxnykiT1lCEvSVJPGfKSJPWUIS9JUk8Z8pIk9ZQhL0lSTxnykiT1lCEvSVJPGfKSJPWUIS9JUk8Z8pIk9ZQhL0lSTxnykiT1lCEvSVJPGfKSJPXUyJBPskeSS5Jck+TqJL/b6rskWZ7kuvZx51ZPklOTrExyZZLnD4y1qPW/Lsmigfp+Sa5qx5yaJNPNIUmSRhvnTP4B4A+qai9gPnBckr2A44GLqmoucFHbBzgEmNsei4HToAts4ATgAGB/4ISB0D4NOHbguAWtPtUckiRphJEhX1U3VtXX2/YPgWuBWcBCYEnrtgQ4rG0vBM6uzqXATkmeBhwMLK+q9VV1K7AcWNDadqiqS6uqgLMnjTVsDkmSNMIGvSefZA7wC8BlwO5VdWNrugnYvW3PAlYNHLa61aarrx5SZ5o5Jq9rcZIVSVasXbt2Q56SJEm9NXbIJ3ky8HfA71XVHYNt7Qy8NvHaHmK6Oarq9KqaV1XzZs6cuTmXIUnSo8ZYIZ/k8XQB//Gq+nQr39wutdM+3tLqa4A9Bg6f3WrT1WcPqU83hyRJGmGcu+sDnAFcW1UfGGhaCkzcIb8IOH+gflS7y34+cHu75L4MOCjJzu2Gu4OAZa3tjiTz21xHTRpr2BySJGmEGWP0eRHwm8BVSa5otXcCJwPnJTkGuAE4vLVdABwKrATuAo4GqKr1Sd4LXN76nVhV69v2m4CzgO2Bz7cH08whSZJGGBnyVfVlIFM0HzikfwHHTTHWmcCZQ+orgL2H1NcNm0OSJI3mb7yTJKmnDHlJknrKkJckqacMeUmSesqQlySppwx5SZJ6ypCXJKmnDHlJknrKkJckqacMeUmSesqQlySppwx5SZJ6ypCXJKmnDHlJknrKkJckqacMeUmSesqQlySpp0aGfJIzk9yS5FsDtV2SLE9yXfu4c6snyalJVia5MsnzB45Z1Ppfl2TRQH2/JFe1Y05NkunmkCRJ4xnnTP4sYMGk2vHARVU1F7io7QMcAsxtj8XAadAFNnACcACwP3DCQGifBhw7cNyCEXNIkqQxjAz5qvonYP2k8kJgSdteAhw2UD+7OpcCOyV5GnAwsLyq1lfVrcByYEFr26GqLq2qAs6eNNawOSRJ0hg29j353avqxrZ9E7B7254FrBrot7rVpquvHlKfbo6fkmRxkhVJVqxdu3Yjno4kSf3zsG+8a2fgtQnWstFzVNXpVTWvqubNnDlzcy5FkqRHjY0N+ZvbpXbax1tafQ2wx0C/2a02XX32kPp0c0iSpDFsbMgvBSbukF8EnD9QP6rdZT8fuL1dcl8GHJRk53bD3UHAstZ2R5L57a76oyaNNWwOSZI0hhmjOiT5BPAyYLckq+nukj8ZOC/JMcANwOGt+wXAocBK4C7gaICqWp/kvcDlrd+JVTVxM9+b6O7g3x74fHswzRySJGkMI0O+ql43RdOBQ/oWcNwU45wJnDmkvgLYe0h93bA5JEnSePyNd5Ik9ZQhL0lSTxnykiT1lCEvSVJPGfKSJPWUIS9JUk8Z8pIk9ZQhL0lSTxnykiT1lCEvSVJPGfKSJPWUIS9JUk8Z8pIk9ZQhL0lSTxnykiT1lCEvSVJPGfKSJPXUVh/ySRYk+XaSlUmO39LrkSTp0WKrDvkk2wAfBg4B9gJel2SvLbsqSZIeHbbqkAf2B1ZW1fVVdR9wDrBwC69JkqRHhRlbegEjzAJWDeyvBg6Y3CnJYmBx270zybcfgbVp09oN+MGWXkTf5f1begXayvl9+AjYDN+Hz5yqYWsP+bFU1enA6Vt6Hdp4SVZU1bwtvQ7psczvw/7Z2i/XrwH2GNif3WqSJGmErT3kLwfmJtkzybbAEcDSLbwmSZIeFbbqy/VV9UCSNwPLgG2AM6vq6i28LG0evt0ibXl+H/ZMqmpLr0GSJG0GW/vlekmStJEMeUmSesqQ18OS5MEkVyT5VpJPJnniBh7/9CSfatv7Jjl0oO1X/VXG0k9LUkn+YmD/D5O8ezPM885J+/+8qefQ5mXI6+G6u6r2raq9gfuAN27IwVX1/ap6ddvdFzh0oG1pVZ286ZYq9ca9wK8n2W0zz/OQkK+qX9zM82kTM+S1KX0JeFaSXZL8fZIrk1ya5HkASV7azvqvSPKNJE9JMqddBdgWOBF4bWt/bZLXJ/nLJDsmuSHJ49o4T0qyKsnjk/xskguTfC3Jl5I8Zws+f+mR8gDdnfC/P7khycwkf5fk8vZ40UB9eZKrk3ysfU/t1tr+vn0PXd1+gyhJTga2b9+PH2+1O9vHc5K8cmDOs5K8Osk2Sf5nm/fKJG/Y7J8JTcuQ1yaRZAbdHxK6CngP8I2qeh7dmcDZrdsfAsdV1b7AS4C7J45vf5vgXcC57crAuQNttwNXAC9tpVcBy6rqfrr/6N5SVfu18T+y+Z6ltFX5MHBkkh0n1U8BPlhVLwD+K/CxVj8BuLiqngt8CnjGwDG/1b6H5gFvTbJrVR3PT67UHTlpjnOBwwHaC/QDgc8BxwC3t7lfABybZM9N9Hy1Ebbqn5PXo8L2Sa5o218CzgAuo/vPhaq6OMmuSXYAvgJ8oJ0VfLqqVicZd55zgdcCl9D9UqSPJHky8IvAJwfGecImeE7SVq+q7khyNvBWBl4wA68A9hr4ntihfa+8GPi1duyFSW4dOOatSX6tbe8BzAXWTTP954FTkjwBWAD8U1XdneQg4HlJJt6C27GN9Z2NfZ56eAx5PVx3tzPzH5squKvq5CSfo3vf/StJDgbuGXOepcCfJdkF2A+4GHgScNvk+aXHkA8BXwf+aqD2OGB+VT3ke2uq78skL6N7YfDCqroryReB7aabtKruaf0Opnvxfc7EcHRX1pZt6BPR5uHlem0OXwKOhB//B/KDdtbxs1V1VVW9n+5XFk9+//yHwFOGDVhVd7ZjTgE+W1UPVtUdwHeSvKbNlST7bJZnJG2Fqmo9cB7dZfIJXwDeMrGTZOJF8Ff4ySX2g4CdW31H4NYW8M8B5g+MdX+Sx08x/bnA0XRvvV3YasuA35k4JsnPJXnSRj49bQKGvDaHdwP7JbkSOBlY1Oq/126yuxK4n+6S36BL6C4zXpHktUPGPRf4jfZxwpHAMUm+CVwNLNx0T0N6VPgLuj8RO+GtwLx249s1/OQnXt4DHJTkW8BrgJvoXlhfCMxIci3d9+ulA2OdDlw5cePdJF+gu0/mH9s9NdC9/38N8PU2z//BK8ZblL/WVpIeA9r75w+2vwnyQuA03+rqP19hSdJjwzOA89qPot4HHLuF16NHgGfykiT1lO/JS5LUU4a8JEk9ZchLktRThrwkST1lyEuS1FP/H4iH2YD19PURAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 576x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-GSdPRPpFm7k",
        "colab_type": "code",
        "outputId": "e83e8d24-87f7-49de-99f6-8b1333b43255",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 215
        }
      },
      "source": [
        "df.text.head(10)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0    @switchfoot http://twitpic.com/2y1zl - Awww, t...\n",
              "1    is upset that he can't update his Facebook by ...\n",
              "2    @Kenichan I dived many times for the ball. Man...\n",
              "3      my whole body feels itchy and like its on fire \n",
              "4    @nationwideclass no, it's not behaving at all....\n",
              "5                        @Kwesidei not the whole crew \n",
              "6                                          Need a hug \n",
              "7    @LOLTrish hey  long time no see! Yes.. Rains a...\n",
              "8                 @Tatiana_K nope they didn't have it \n",
              "9                            @twittera que me muera ? \n",
              "Name: text, dtype: object"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 10
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "t00MPV91M7Zw",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "stop_words = stopwords.words('english')\n",
        "stemmer = SnowballStemmer('english')\n",
        "\n",
        "text_cleaning_re = \"@\\S+|https?:\\S+|http?:\\S|[^A-Za-z0-9]+\""
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Izdo8XuEQ0Tu",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def preprocess(text, stem=False):\n",
        "  text = re.sub(text_cleaning_re, ' ', str(text).lower()).strip()\n",
        "  tokens = []\n",
        "  for token in text.split():\n",
        "    if token not in stop_words:\n",
        "      if stem:\n",
        "        tokens.append(stemmer.stem(token))\n",
        "      else:\n",
        "        tokens.append(token)\n",
        "  return \" \".join(tokens)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MX-DPP9iSD-O",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "df.text = df.text.apply(lambda x: preprocess(x))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cA4R-9wmbWhc",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from wordcloud import WordCloud\n",
        "plt.figure(figsize = (20,20)) \n",
        "wc = WordCloud(max_words = 2000 , width = 1600 , height = 800).generate(\" \".join(df[df.sentiment == 'Positive'].text))\n",
        "plt.imshow(wc , interpolation = 'bilinear')"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "A_liO0YQlXyV",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "plt.figure(figsize = (20,20)) \n",
        "wc = WordCloud(max_words = 2000 , width = 1600 , height = 800).generate(\" \".join(df[df.sentiment == 'Negative'].text))\n",
        "plt.imshow(wc , interpolation = 'bilinear')"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FiPUt--2S3K1",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "TRAIN_SIZE = 0.8\n",
        "MAX_NB_WORDS = 100000\n",
        "MAX_SEQUENCE_LENGTH = 30\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BNcfdCWmSLGO",
        "colab_type": "code",
        "outputId": "4be263d9-cbd2-4a34-daff-bbad6000a726",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 53
        }
      },
      "source": [
        "train_data, test_data = train_test_split(df, test_size=1-TRAIN_SIZE,\n",
        "                                         random_state=7)\n",
        "print(\"Train Data size:\", len(train_data))\n",
        "print(\"Test Data size\", len(test_data))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train Data size: 1280000\n",
            "Test Data size 320000\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "R3lYTX7hGu2z",
        "colab_type": "code",
        "outputId": "90bd571f-63dc-4a75-ad18-3d0fc3ac2dbd",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 359
        }
      },
      "source": [
        "train_data.head(10)"
      ],
      "execution_count": 0,
      "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>sentiment</th>\n",
              "      <th>text</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>23786</th>\n",
              "      <td>Negative</td>\n",
              "      <td>need friends</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>182699</th>\n",
              "      <td>Negative</td>\n",
              "      <td>im trying call impossible</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>476661</th>\n",
              "      <td>Negative</td>\n",
              "      <td>good pace going 3k 13 min missed 5k turn ended...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1181490</th>\n",
              "      <td>Positive</td>\n",
              "      <td>u gonna shows ny soon luv see u live</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>878773</th>\n",
              "      <td>Positive</td>\n",
              "      <td>hell yea get em tattoos ink free wish parents ...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>130866</th>\n",
              "      <td>Negative</td>\n",
              "      <td>yeah need 2 see ur mom calls back first rememb...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1235876</th>\n",
              "      <td>Positive</td>\n",
              "      <td>sounds like cup tea sign</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>717314</th>\n",
              "      <td>Negative</td>\n",
              "      <td>tired want sleep wtf</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>969880</th>\n",
              "      <td>Positive</td>\n",
              "      <td>amazing wish</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>748698</th>\n",
              "      <td>Negative</td>\n",
              "      <td>thank god wkrn abc affiliate nashville back mi...</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "        sentiment                                               text\n",
              "23786    Negative                                       need friends\n",
              "182699   Negative                          im trying call impossible\n",
              "476661   Negative  good pace going 3k 13 min missed 5k turn ended...\n",
              "1181490  Positive               u gonna shows ny soon luv see u live\n",
              "878773   Positive  hell yea get em tattoos ink free wish parents ...\n",
              "130866   Negative  yeah need 2 see ur mom calls back first rememb...\n",
              "1235876  Positive                           sounds like cup tea sign\n",
              "717314   Negative                               tired want sleep wtf\n",
              "969880   Positive                                       amazing wish\n",
              "748698   Negative  thank god wkrn abc affiliate nashville back mi..."
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 17
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "aGAQ7I04G8Rg",
        "colab_type": "code",
        "outputId": "db90c31b-f307-49c1-e5c7-5f9ae53b7d95",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 197
        }
      },
      "source": [
        "print(\"Train_Data Distribution: \\n\",train_data.sentiment.value_counts())\n",
        "print(\"\\n\\nTest Data Distribution: \\n\", test_data.sentiment.value_counts())"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train_Data Distribution: \n",
            " Positive    640542\n",
            "Negative    639458\n",
            "Name: sentiment, dtype: int64\n",
            "\n",
            "\n",
            "Test Data Distribution: \n",
            " Negative    160542\n",
            "Positive    159458\n",
            "Name: sentiment, dtype: int64\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "uuxGOkxuTIoX",
        "colab_type": "code",
        "outputId": "4ccac2e8-4910-4002-f059-32dc1df447f0",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        }
      },
      "source": [
        "tokenizer = Tokenizer()\n",
        "tokenizer.fit_on_texts(train_data.text)\n",
        "\n",
        "word_index = tokenizer.word_index\n",
        "vocab_size = len(tokenizer.word_index) + 1\n",
        "print(\"Vocabulary Size :\", vocab_size)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Vocabulary Size : 290575\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "aM1bHdCXWMRz",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "x_train = pad_sequences(tokenizer.texts_to_sequences(train_data.text),\n",
        "                        maxlen = MAX_SEQUENCE_LENGTH)\n",
        "x_test = pad_sequences(tokenizer.texts_to_sequences(test_data.text),\n",
        "                       maxlen = MAX_SEQUENCE_LENGTH)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XaLUJukOILcE",
        "colab_type": "code",
        "outputId": "440a0aeb-4725-4580-89d2-869abbcfb9c6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 53
        }
      },
      "source": [
        "print(\"Training X Shape:\",x_train.shape)\n",
        "print(\"Testing X Shape:\",x_test.shape)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Training X Shape: (1280000, 30)\n",
            "Testing X Shape: (320000, 30)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "oLo5VGWeWxdn",
        "colab_type": "code",
        "outputId": "3382ee27-d765-46d9-b735-edee69933ff8",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        }
      },
      "source": [
        "labels = train_data.sentiment.unique().tolist()\n",
        "labels.append('Neutral')\n",
        "labels"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['Negative', 'Positive', 'Neutral']"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 22
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nhQEUL5RXJoH",
        "colab_type": "code",
        "outputId": "23273a1d-5487-4a93-d69d-a08c061709b6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 53
        }
      },
      "source": [
        "encoder = LabelEncoder()\n",
        "encoder.fit(train_data.sentiment.to_list())\n",
        "\n",
        "y_train = encoder.transform(train_data.sentiment.to_list())\n",
        "y_test = encoder.transform(test_data.sentiment.to_list())\n",
        "\n",
        "y_train = y_train.reshape(-1,1)\n",
        "y_test = y_test.reshape(-1,1)\n",
        "\n",
        "print(\"y_train shape:\", y_train.shape)\n",
        "print(\"y_test shape:\", y_test.shape)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "y_train shape: (1280000, 1)\n",
            "y_test shape: (320000, 1)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gGK228HaX_IY",
        "colab_type": "code",
        "outputId": "3f9a5228-fcf5-477a-ec95-815fa45843f2",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 467
        }
      },
      "source": [
        "!wget http://nlp.stanford.edu/data/glove.6B.zip\n",
        "!unzip glove.6B.zip"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "--2020-05-13 09:07:13--  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",
            "--2020-05-13 09:07:13--  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",
            "--2020-05-13 09:07:13--  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: ‘glove.6B.zip’\n",
            "\n",
            "glove.6B.zip        100%[===================>] 822.24M  2.00MB/s    in 6m 30s  \n",
            "\n",
            "2020-05-13 09:13:43 (2.11 MB/s) - ‘glove.6B.zip’ saved [862182613/862182613]\n",
            "\n",
            "Archive:  glove.6B.zip\n",
            "  inflating: glove.6B.50d.txt        \n",
            "  inflating: glove.6B.100d.txt       \n",
            "  inflating: glove.6B.200d.txt       \n",
            "  inflating: glove.6B.300d.txt       \n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Ha-7BByj0mv4",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "GLOVE_EMB = '/content/glove.6B.300d.txt'\n",
        "EMBEDDING_DIM = 300\n",
        "LR = 1e-3\n",
        "BATCH_SIZE = 2056\n",
        "EPOCHS = 20\n",
        "MODEL_PATH = '/content/drive/My Drive/Dev/sentiment-classification/10_05_best_model.hdf5'"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jtxwnII_yfg8",
        "colab_type": "code",
        "outputId": "94d24028-2a55-4b1e-9664-d172ad8b1e3d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        }
      },
      "source": [
        "embeddings_index = {}\n",
        "\n",
        "f = open(GLOVE_EMB)\n",
        "for line in f:\n",
        "  values = line.split()\n",
        "  word = value = values[0]\n",
        "  coefs = np.asarray(values[1:], dtype='float32')\n",
        "  embeddings_index[word] = coefs\n",
        "f.close()\n",
        "\n",
        "print('Found %s word vectors.' %len(embeddings_index))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Found 400000 word vectors.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KVMsB1ps1OFS",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "embedding_matrix = np.zeros((vocab_size, EMBEDDING_DIM))\n",
        "for word, i in word_index.items():\n",
        "  embedding_vector = embeddings_index.get(word)\n",
        "  if embedding_vector is not None:\n",
        "    embedding_matrix[i] = embedding_vector"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2Z527v-y2DLG",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "embedding_layer = tf.keras.layers.Embedding(vocab_size,\n",
        "                                          EMBEDDING_DIM,\n",
        "                                          weights=[embedding_matrix],\n",
        "                                          input_length=MAX_SEQUENCE_LENGTH,\n",
        "                                          trainable=False)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "G0hA8RSg4-1G",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from tensorflow.keras.layers import Conv1D, Bidirectional, LSTM, Dense, Input, Dropout\n",
        "from tensorflow.keras.layers import SpatialDropout1D\n",
        "from tensorflow.keras.callbacks import ModelCheckpoint"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VV3ay2--3RHP",
        "colab_type": "code",
        "outputId": "5fc5f2c0-f98c-43c4-a0ac-732f55bf9493",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 91
        }
      },
      "source": [
        "sequence_input = Input(shape=(MAX_SEQUENCE_LENGTH,), dtype='int32')\n",
        "embedding_sequences = embedding_layer(sequence_input)\n",
        "x = SpatialDropout1D(0.2)(embedding_sequences)\n",
        "x = Conv1D(64, 5, activation='relu')(x)\n",
        "x = Bidirectional(LSTM(64, dropout=0.2, recurrent_dropout=0.2))(x)\n",
        "x = Dense(512, activation='relu')(x)\n",
        "x = Dropout(0.5)(x)\n",
        "x = Dense(512, activation='relu')(x)\n",
        "outputs = Dense(1, activation='sigmoid')(x)\n",
        "model = tf.keras.Model(sequence_input, outputs)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Layer lstm_2 will not use cuDNN kernel since it doesn't meet the cuDNN kernel criteria. It will use generic GPU kernel as fallback when running on GPU\n",
            "WARNING:tensorflow:Layer lstm_2 will not use cuDNN kernel since it doesn't meet the cuDNN kernel criteria. It will use generic GPU kernel as fallback when running on GPU\n",
            "WARNING:tensorflow:Layer lstm_2 will not use cuDNN kernel since it doesn't meet the cuDNN kernel criteria. It will use generic GPU kernel as fallback when running on GPU\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9z3Y6eMpuAUS",
        "colab_type": "code",
        "outputId": "4e34ca11-a7ac-4a98-d6cc-73dac5fdc341",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 242
        }
      },
      "source": [
        "from tensorflow.keras.optimizers import Adam\n",
        "from tensorflow.keras.callbacks import ReduceLROnPlateau\n",
        "model.compile(optimizer=Adam(learning_rate=LR), loss='binary_crossentropy',\n",
        "              metrics=['accuracy'])\n",
        "def scheduler(epoch):\n",
        "  if epoch<10:\n",
        "    return LR\n",
        "  else:\n",
        "    return LR * tf.math.exp(0.1 * (10 - epoch))\n",
        "\n",
        "checkpoint = ModelCheckpoint(filepath=MODEL_PATH,\n",
        "                             save_best_only=True,\n",
        "                             mode='min',\n",
        "                             verbose=1,\n",
        "                             monitor='val_loss')\n",
        "\n",
        "LRscheduler = tf.keras.callbacks.LearningRateScheduler(scheduler)\n",
        "    "
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "error",
          "ename": "NameError",
          "evalue": "ignored",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-1-d45f35c1f604>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0mtensorflow\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mkeras\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0moptimizers\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mAdam\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      2\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0mtensorflow\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mkeras\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcallbacks\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mReduceLROnPlateau\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 3\u001b[0;31m model.compile(optimizer=Adam(learning_rate=LR), loss='binary_crossentropy',\n\u001b[0m\u001b[1;32m      4\u001b[0m               metrics=['accuracy'])\n\u001b[1;32m      5\u001b[0m \u001b[0;32mdef\u001b[0m \u001b[0mscheduler\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mepoch\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mNameError\u001b[0m: name 'model' is not defined"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "i-dp7gaO5qJy",
        "colab_type": "code",
        "outputId": "b4990cf6-9424-4f72-a768-b91b1eab43d6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 629
        }
      },
      "source": [
        "history = model.fit(x_train, y_train, batch_size=BATCH_SIZE, epochs=EPOCHS,\n",
        "                    validation_data=(x_test, y_test), callbacks=[checkpoint, LRscheduler])"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/20\n",
            "623/623 [==============================] - ETA: 0s - loss: 0.5256 - accuracy: 0.7334\n",
            "Epoch 00001: val_loss improved from inf to 0.48764, saving model to /content/drive/My Drive/Dev/sentiment-classification/10_05_best_model.hdf5\n",
            "623/623 [==============================] - 178s 286ms/step - loss: 0.5256 - accuracy: 0.7334 - val_loss: 0.4876 - val_accuracy: 0.7621 - lr: 0.0010\n",
            "Epoch 2/20\n",
            "623/623 [==============================] - ETA: 0s - loss: 0.4915 - accuracy: 0.7595\n",
            "Epoch 00002: val_loss improved from 0.48764 to 0.47662, saving model to /content/drive/My Drive/Dev/sentiment-classification/10_05_best_model.hdf5\n",
            "623/623 [==============================] - 180s 289ms/step - loss: 0.4915 - accuracy: 0.7595 - val_loss: 0.4766 - val_accuracy: 0.7704 - lr: 0.0010\n",
            "Epoch 3/20\n",
            "623/623 [==============================] - ETA: 0s - loss: 0.4806 - accuracy: 0.7665\n",
            "Epoch 00003: val_loss improved from 0.47662 to 0.46903, saving model to /content/drive/My Drive/Dev/sentiment-classification/10_05_best_model.hdf5\n",
            "623/623 [==============================] - 179s 288ms/step - loss: 0.4806 - accuracy: 0.7665 - val_loss: 0.4690 - val_accuracy: 0.7754 - lr: 0.0010\n",
            "Epoch 4/20\n",
            "623/623 [==============================] - ETA: 0s - loss: 0.4743 - accuracy: 0.7709\n",
            "Epoch 00004: val_loss improved from 0.46903 to 0.46730, saving model to /content/drive/My Drive/Dev/sentiment-classification/10_05_best_model.hdf5\n",
            "623/623 [==============================] - 181s 290ms/step - loss: 0.4743 - accuracy: 0.7709 - val_loss: 0.4673 - val_accuracy: 0.7763 - lr: 0.0010\n",
            "Epoch 5/20\n",
            "623/623 [==============================] - ETA: 0s - loss: 0.4693 - accuracy: 0.7739\n",
            "Epoch 00005: val_loss improved from 0.46730 to 0.46309, saving model to /content/drive/My Drive/Dev/sentiment-classification/10_05_best_model.hdf5\n",
            "623/623 [==============================] - 179s 287ms/step - loss: 0.4693 - accuracy: 0.7739 - val_loss: 0.4631 - val_accuracy: 0.7784 - lr: 0.0010\n",
            "Epoch 6/20\n",
            "623/623 [==============================] - ETA: 0s - loss: 0.4659 - accuracy: 0.7761\n",
            "Epoch 00006: val_loss improved from 0.46309 to 0.46126, saving model to /content/drive/My Drive/Dev/sentiment-classification/10_05_best_model.hdf5\n",
            "623/623 [==============================] - 183s 294ms/step - loss: 0.4659 - accuracy: 0.7761 - val_loss: 0.4613 - val_accuracy: 0.7795 - lr: 0.0010\n",
            "Epoch 7/20\n",
            "623/623 [==============================] - ETA: 0s - loss: 0.4628 - accuracy: 0.7783\n",
            "Epoch 00007: val_loss did not improve from 0.46126\n",
            "623/623 [==============================] - 176s 283ms/step - loss: 0.4628 - accuracy: 0.7783 - val_loss: 0.4617 - val_accuracy: 0.7795 - lr: 0.0010\n",
            "Epoch 8/20\n",
            "623/623 [==============================] - ETA: 0s - loss: 0.4609 - accuracy: 0.7794\n",
            "Epoch 00008: val_loss improved from 0.46126 to 0.45998, saving model to /content/drive/My Drive/Dev/sentiment-classification/10_05_best_model.hdf5\n",
            "623/623 [==============================] - 180s 290ms/step - loss: 0.4609 - accuracy: 0.7794 - val_loss: 0.4600 - val_accuracy: 0.7801 - lr: 0.0010\n",
            "Epoch 9/20\n",
            "257/623 [===========>..................] - ETA: 1:40 - loss: 0.4575 - accuracy: 0.7809"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kBucfSHqhjDb",
        "colab_type": "code",
        "outputId": "39d3fc30-7400-490f-ab38-902cbfe8552e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 313
        }
      },
      "source": [
        "plt.plot(history.history['accuracy'], c= 'b')\n",
        "plt.plot(history.history['val_accuracy'], c='r')\n",
        "plt.title('model accuracy')\n",
        "plt.ylabel('accuracy')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['LSTM_train', 'LSTM_val'], loc='upper left')"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.legend.Legend at 0x7f55942b9b00>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 36
        },
        {
          "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": "NZ98h1JKsZlw",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import pickle \n",
        "#saving Tokenizer() object for Text Tokenizer on later use and deployment\n",
        "TOKENIZER_PATH = '/content/drive/My Drive/Dev/sentiment-classification/tokenizer_obj.pickle'\n",
        "with open(TOKENIZER_PATH, 'wb') as handle:\n",
        "  pickle.dump(tokenizer, handle, protocol=pickle.HIGHEST_PROTOCOL)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EyuPcnAo7rxF",
        "colab_type": "code",
        "outputId": "4aac6fdc-5ff9-45df-967e-c242091884a9",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 91
        }
      },
      "source": [
        "with open(TOKENIZER_PATH,'rb') as handle:\n",
        "  tokenizer = pickle.load(handle)\n",
        "\n",
        "model = tf.keras.models.load_model(MODEL_PATH)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:Layer lstm_1 will not use cuDNN kernel since it doesn't meet the cuDNN kernel criteria. It will use generic GPU kernel as fallback when running on GPU\n",
            "WARNING:tensorflow:Layer lstm_1 will not use cuDNN kernel since it doesn't meet the cuDNN kernel criteria. It will use generic GPU kernel as fallback when running on GPU\n",
            "WARNING:tensorflow:Layer lstm_1 will not use cuDNN kernel since it doesn't meet the cuDNN kernel criteria. It will use generic GPU kernel as fallback when running on GPU\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lbZ161ewagmQ",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "SENTIMENT_THRESHOLDS = (0.4, 0.7)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ysVbRFm1XzbY",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def decode_sentiment(score, include_neutral=True):\n",
        "    if include_neutral:        \n",
        "        label = 'Neutral'\n",
        "        if score <= SENTIMENT_THRESHOLDS[0]:\n",
        "            label = 'Negative'\n",
        "        elif score >= SENTIMENT_THRESHOLDS[1]:\n",
        "            label = \"Positive\"\n",
        "\n",
        "        return label\n",
        "    else:\n",
        "        return 'Negative' if score < 0.5 else 'Positive'"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-EK0O7VhZzi4",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def predict(text, include_neutral=True):\n",
        "    # Tokenize text\n",
        "    x_test = pad_sequences(tokenizer.texts_to_sequences([text]), maxlen=MAX_SEQUENCE_LENGTH)\n",
        "    # Predict\n",
        "    score = model.predict([x_test])[0]\n",
        "    # Decode sentiment\n",
        "    label = decode_sentiment(score, include_neutral=include_neutral)\n",
        "\n",
        "    return {\"label\": label, \"score\": float(score)}  "
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vy1F4-gDYAut",
        "colab_type": "code",
        "outputId": "a847348b-a370-4c3f-9fdf-a2e8af3ffe41",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        }
      },
      "source": [
        "scores = model.predict(x_test, verbose=1, batch_size=10000)\n",
        "y_pred_1d = [decode_sentiment(score, include_neutral=False) for score in scores]"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "32/32 [==============================] - 12s 369ms/step\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fSdW1PqnW7G6",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import itertools\n",
        "from sklearn.metrics import confusion_matrix, classification_report, accuracy_score\n",
        "def plot_confusion_matrix(cm, classes,\n",
        "                          title='Confusion matrix',\n",
        "                          cmap=plt.cm.Blues):\n",
        "    \"\"\"\n",
        "    This function prints and plots the confusion matrix.\n",
        "    Normalization can be applied by setting `normalize=True`.\n",
        "    \"\"\"\n",
        "\n",
        "    cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]\n",
        "\n",
        "    plt.imshow(cm, interpolation='nearest', cmap=cmap)\n",
        "    plt.title(title, fontsize=20)\n",
        "    plt.colorbar()\n",
        "    tick_marks = np.arange(len(classes))\n",
        "    plt.xticks(tick_marks, classes, fontsize=13)\n",
        "    plt.yticks(tick_marks, classes, fontsize=13)\n",
        "\n",
        "    fmt = '.2f'\n",
        "    thresh = cm.max() / 2.\n",
        "    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
        "        plt.text(j, i, format(cm[i, j], fmt),\n",
        "                 horizontalalignment=\"center\",\n",
        "                 color=\"white\" if cm[i, j] > thresh else \"black\")\n",
        "\n",
        "    plt.ylabel('True label', fontsize=17)\n",
        "    plt.xlabel('Predicted label', fontsize=17)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "w5ZaUjUfW8En",
        "colab_type": "code",
        "outputId": "045b4737-9419-4c8e-f6c2-d4f8fc5322e3",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 370
        }
      },
      "source": [
        "cnf_matrix = confusion_matrix(test_data.sentiment.to_list(), y_pred_1d)\n",
        "plt.figure(figsize=(6,6))\n",
        "plot_confusion_matrix(cnf_matrix, classes=test_data.sentiment.unique(), title=\"Confusion matrix\")\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x432 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kWOsRkur9xEZ",
        "colab_type": "code",
        "outputId": "415ed819-466e-452f-c07b-c08c883f50af",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 179
        }
      },
      "source": [
        "print(classification_report(list(test_data.sentiment), y_pred_1d))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "    Negative       0.79      0.77      0.78    160542\n",
            "    Positive       0.78      0.80      0.79    159458\n",
            "\n",
            "    accuracy                           0.78    320000\n",
            "   macro avg       0.78      0.78      0.78    320000\n",
            "weighted avg       0.78      0.78      0.78    320000\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    }
  ]
}