sentiment_analysis.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.7.4-final"
    },
    "orig_nbformat": 2,
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3",
      "language": "python"
    },
    "colab": {
      "name": "sentiment_analysis.ipynb",
      "provenance": [],
      "collapsed_sections": []
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "wyHF8G-8rlUz"
      },
      "source": [
        "import pandas as pd\n",
        "import numpy as np\n",
        "\n",
        "PATH = './drive/MyDrive/Colab Notebooks/Datasets/training.1600000.processed.noemoticon.csv'\n",
        "# data from AI crowd: https://www.aicrowd.com/challenges/epfl-ml-text-classification\n",
        "PATH_POS = './drive/MyDrive/Colab Notebooks/Datasets/train_pos_full.txt'\n",
        "PATH_NEG = './drive/MyDrive/Colab Notebooks/Datasets/train_neg_full.txt'\n",
        "PICKLE_PATH = './backup_sentiment'"
      ],
      "execution_count": 4,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "M1GOLOxGu8f_",
        "outputId": "dc8b3ae0-6980-4d97-d888-f78c9cb3399c"
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ],
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Mounted at /content/drive\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vlNxZuKSrlU0"
      },
      "source": [
        "# function to read the .txt file of the AI crowd data set\n",
        "def get_new_tweets():\n",
        "    pos= list()\n",
        "    neg = list()\n",
        "    with open(PATH_POS,errors = \"ignore\") as f:\n",
        "        for line in f:\n",
        "            pos.append(line.replace(\"\\n\",\"\"))\n",
        "    with open(PATH_NEG,errors = \"ignore\") as f:\n",
        "        for line in f:\n",
        "            neg.append(line.replace(\"\\n\",\"\"))\n",
        "    return pd.Series(np.array(pos)),pd.Series(np.array(neg))"
      ],
      "execution_count": 6,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1LzesEncrlU1"
      },
      "source": [
        "tweet_pos,tweet_neg = get_new_tweets()"
      ],
      "execution_count": 7,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Ux55onG1rlU2",
        "outputId": "d7a59485-9b12-405b-92a7-e7c56da29ad8"
      },
      "source": [
        "# importing the kaggle dataset\n",
        "try:  \n",
        "    tweets = pd.read_pickle(PICKLE_PATH)\n",
        "except:\n",
        "    header_list = [\"target\", \"id\", \"date\" ,\"flag\", \"user\", \"text\"]\n",
        "    print(\"there is no backup\")\n",
        "    tweets = pd.read_csv(PATH, encoding = \"ISO-8859-1\", names=header_list)\n",
        "    tweets.to_pickle(PICKLE_PATH)\n",
        "    # tweet_pos,tweet_neg = get_new_tweets()"
      ],
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "there is no backup\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 371
        },
        "id": "DGkVKMb_rlU3",
        "outputId": "abb07115-b5d9-42f1-f2f1-bd982d5e4dc1"
      },
      "source": [
        "sentiment_num2name = {\n",
        "     0: \"Negative\",\n",
        "     2: \"Neutral\",\n",
        "     4: \"Positive\",\n",
        "}\n",
        "tweets[\"target\"] = tweets[\"target\"].apply(lambda num: sentiment_num2name[num])\n",
        "tweets.head()"
      ],
      "execution_count": 60,
      "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>target</th>\n",
              "      <th>id</th>\n",
              "      <th>date</th>\n",
              "      <th>flag</th>\n",
              "      <th>user</th>\n",
              "      <th>text</th>\n",
              "      <th>month</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>Negative</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",
              "      <td>4</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>Negative</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",
              "      <td>4</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>Negative</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",
              "      <td>4</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>Negative</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",
              "      <td>4</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>Negative</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",
              "      <td>4</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "     target  ...  month\n",
              "0  Negative  ...      4\n",
              "1  Negative  ...      4\n",
              "2  Negative  ...      4\n",
              "3  Negative  ...      4\n",
              "4  Negative  ...      4\n",
              "\n",
              "[5 rows x 7 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 60
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 302
        },
        "id": "BaDje1lRSBFT",
        "outputId": "cd0e94d8-758c-4e44-f8fd-6f922c40e38b"
      },
      "source": [
        "tweets.head()"
      ],
      "execution_count": 10,
      "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>target</th>\n",
              "      <th>id</th>\n",
              "      <th>date</th>\n",
              "      <th>flag</th>\n",
              "      <th>user</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": [
              "   target  ...                                               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": 10
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9npWbAEnR3FO"
      },
      "source": [
        ""
      ],
      "execution_count": 11,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "B49V1k6WSSzT"
      },
      "source": [
        "import time\r\n",
        "# adding a column called month with the number of the month\r\n",
        "# this is done by applying an anonymous function to each row\r\n",
        "# using the function time.streptime which converts a string date in the asctime format\r\n",
        "# to a gm time structure\r\n",
        "tweets['month'] = tweets['date'].apply(lambda x: time.strptime(x[0:19] + x[23:]).tm_mon)\r\n",
        "tweets['hour'] = tweets['date'].apply(lambda x: time.strptime(x[0:19] + x[23:]).tm_hour)\r\n",
        "tweets['weekday'] = tweets['date'].apply(lambda x: time.asctime(time.strptime(x[0:19] + x[23:]))[0:3])"
      ],
      "execution_count": 176,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 564
        },
        "id": "StLCrILP9pua",
        "outputId": "152dc228-8cf0-4843-9403-5f6b431c57e1"
      },
      "source": [
        "tweets.head()"
      ],
      "execution_count": 232,
      "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>target</th>\n",
              "      <th>id</th>\n",
              "      <th>date</th>\n",
              "      <th>flag</th>\n",
              "      <th>user</th>\n",
              "      <th>text</th>\n",
              "      <th>month</th>\n",
              "      <th>hour</th>\n",
              "      <th>weekday</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>Negative</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",
              "      <td>4</td>\n",
              "      <td>22</td>\n",
              "      <td>Mon</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>Negative</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",
              "      <td>4</td>\n",
              "      <td>22</td>\n",
              "      <td>Mon</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>Negative</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",
              "      <td>4</td>\n",
              "      <td>22</td>\n",
              "      <td>Mon</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>Negative</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",
              "      <td>4</td>\n",
              "      <td>22</td>\n",
              "      <td>Mon</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>Negative</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",
              "      <td>4</td>\n",
              "      <td>22</td>\n",
              "      <td>Mon</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "     target          id                          date  ... month hour weekday\n",
              "0  Negative  1467810369  Mon Apr 06 22:19:45 PDT 2009  ...     4   22     Mon\n",
              "1  Negative  1467810672  Mon Apr 06 22:19:49 PDT 2009  ...     4   22     Mon\n",
              "2  Negative  1467810917  Mon Apr 06 22:19:53 PDT 2009  ...     4   22     Mon\n",
              "3  Negative  1467811184  Mon Apr 06 22:19:57 PDT 2009  ...     4   22     Mon\n",
              "4  Negative  1467811193  Mon Apr 06 22:19:57 PDT 2009  ...     4   22     Mon\n",
              "\n",
              "[5 rows x 9 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 232
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VuLCn6_1Ux7s"
      },
      "source": [
        "# grouping the tweets per month\r\n",
        "group_pos = tweets[tweets['target']=='Positive'].groupby(['month'])\r\n",
        "group_neg = tweets[tweets['target']=='Negative'].groupby(['month'])\r\n",
        "group_all = tweets.groupby(['month'])"
      ],
      "execution_count": 177,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vg-vDGttVJJU"
      },
      "source": [
        "# counting the number of positive, negative and all tweets in each months\r\n",
        "y_pos = group_pos.count()['id']\r\n",
        "y_neg = group_neg.count()['id']\r\n",
        "y_all = group_all.count()['id']"
      ],
      "execution_count": 178,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 285
        },
        "id": "jT_9IPzBWC8Y",
        "outputId": "9cf9d4c0-0b2f-4c27-b43b-50f72e2b039f"
      },
      "source": [
        "# plotting the number of tweets per months\r\n",
        "from matplotlib import pyplot as plt\r\n",
        "plt.plot(np.arange(4,7), y_all) \r\n",
        "plt.plot(np.arange(4,7), y_pos)\r\n",
        "plt.plot(np.arange(4,7), y_neg)\r\n",
        "plt.legend(['All tweets', 'Positive tweets', 'Negative tweets'])\r\n",
        "plt.xticks([4, 5, 6], ['April', 'Mai', 'June'])\r\n",
        "plt.xlabel('month')\r\n",
        "plt.ylabel('Num of tweets')\r\n",
        "plt.rcParams.update({'font.size': 12})\r\n",
        "plt.savefig('tweets_april_mai_june', dpi = 300)"
      ],
      "execution_count": 179,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nE1p4pdT3F9S"
      },
      "source": [
        "# calculating the number of tweet per hour on each day of the week\r\n",
        "days = ['Sun', 'Mon', 'Tue', 'Wed', 'Thu', 'Fri', 'Sat']\r\n",
        "time = list(np.arange(23))\r\n",
        "num_tweet_all = np.zeros((len(days), len(time)))\r\n",
        "num_tweet_pos = np.zeros((len(days), len(time)))\r\n",
        "num_tweet_neg = np.zeros((len(days), len(time)))\r\n",
        "for i in range(len(days)):\r\n",
        "    # counts the number of occurence of each hour\r\n",
        "    # .value_counts() returns the occurences of hours which need to be reordered by evaluating it at \"time\"\r\n",
        "    num_tweet_all[i] = np.array(tweets[tweets['weekday']==days[i]]['hour'].value_counts()[time])\r\n",
        "    # The following two lines didn't work because the dataset has no positive or negative tweets at some hours\r\n",
        "    # It seems the data set was manipulated to balance the number of positive and negatice tweets\r\n",
        "    # This results in non uniformous distribution of positive and negative tweets in time\r\n",
        "    # num_tweet_pos[i] = np.array(tweets[(tweets['weekday']==days[i]) & (tweets['target']=='Positive')]['hour'].value_counts()[time])\r\n",
        "    # num_tweet_neg[i] = np.array(tweets[(tweets['weekday']==days[i]) & (tweets['target']=='Negative')]['hour'].value_counts()[time])"
      ],
      "execution_count": 235,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 285
        },
        "id": "V39Zj9fn5PpC",
        "outputId": "69e7772c-c3e3-412b-8ff7-f22d06ded492"
      },
      "source": [
        "# plotting the number of tweets per hour on each day of the week\r\n",
        "fig, ax = plt.subplots()\r\n",
        "ax.set_prop_cycle(color = ['red', 'goldenrod', 'green', 'cyan', 'blue', 'purple', 'magenta'])\r\n",
        "plt.plot(time,num_tweet_all.T)\r\n",
        "plt.legend(days)\r\n",
        "plt.ylabel('Number of Tweets')\r\n",
        "plt.xlabel('Hour')\r\n",
        "plt.xticks(np.linspace(0,23,24));\r\n",
        "plt.xlim((0, 30))\r\n",
        "plt.savefig('tweets_per_hour', dpi = 300)"
      ],
      "execution_count": 242,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AoWzZCD8rlU4"
      },
      "source": [
        "# Preprocessing\n",
        "from sklearn.feature_extraction.text import TfidfVectorizer\n",
        "from sklearn.feature_extraction.text import TfidfTransformer\n",
        "from sklearn.model_selection import train_test_split\n",
        "\n",
        "# Building classification models\n",
        "from sklearn.svm import LinearSVC\n",
        "from sklearn.pipeline import Pipeline\n",
        "\n",
        "# Model evaluation\n",
        "from sklearn import metrics\n",
        "from sklearn.metrics import classification_report, confusion_matrix"
      ],
      "execution_count": 61,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "kloTLcBbrlU4",
        "outputId": "f184bb47-a659-435e-db25-ff434cf9bf3d"
      },
      "source": [
        "X_all = tweets[\"text\"]\n",
        "y_all = tweets[\"target\"]\n",
        "len_pos = len(tweet_pos)\n",
        "len_neg = len(tweet_neg)\n",
        "X_all = X_all.append(tweet_pos)\n",
        "X_all =X_all.append(tweet_neg)\n",
        "y_all =y_all.append(pd.Series(np.array(['Positive']*len_pos)))\n",
        "y_all =y_all.append(pd.Series(np.array(['Negative']*len_neg)))\n",
        "y_all"
      ],
      "execution_count": 63,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0          Negative\n",
              "1          Negative\n",
              "2          Negative\n",
              "3          Negative\n",
              "4          Negative\n",
              "             ...   \n",
              "1249995    Negative\n",
              "1249996    Negative\n",
              "1249997    Negative\n",
              "1249998    Negative\n",
              "1249999    Negative\n",
              "Length: 4100000, dtype: object"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 63
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HwljsGWIrlU5"
      },
      "source": [
        "\n",
        "X_train, X_test, y_train, y_test = train_test_split(X_all, y_all, test_size=0.25, random_state=42)\n",
        "\n",
        "X_train, X_valid, y_train, y_valid = train_test_split(X_train, y_train, test_size=0.1, random_state=42)"
      ],
      "execution_count": 64,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "wbl1zYZSrlU5",
        "outputId": "ebbed825-83be-49c4-d6a2-e5a06e10fb7c"
      },
      "source": [
        "tfidf = TfidfVectorizer()\n",
        "tfidf.fit_transform(X_train)"
      ],
      "execution_count": 65,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<2767500x810276 sparse matrix of type '<class 'numpy.float64'>'\n",
              "\twith 32696184 stored elements in Compressed Sparse Row format>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 65
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Hjdge-6grlU6"
      },
      "source": [
        "def train(tfidf, model, train_data, train_labels, test_data):\n",
        "    model.fit(tfidf.transform(train_data), train_labels)\n",
        "    preds = model.predict(tfidf.transform(test_data))\n",
        "    \n",
        "    return preds"
      ],
      "execution_count": 66,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xKnYXD2GrlU6"
      },
      "source": [
        "def grade(model, preds, test_labels):\n",
        "    print(metrics.classification_report(test_labels, preds))\n",
        "    \n",
        "    cm = confusion_matrix(test_labels, preds)\n",
        "    cm_normd = cm / cm.sum(axis=1).reshape(-1, 1)\n",
        "    \n",
        "    heatmap_kwargs = dict(\n",
        "        cmap=\"YlGnBu\",\n",
        "        xticklabels=model.classes_,\n",
        "        yticklabels=model.classes_,\n",
        "        vmin=0.,\n",
        "        vmax=1.,\n",
        "        annot=True,\n",
        "        annot_kws={\"size\": 10},\n",
        "    )\n",
        "    \n",
        "    sns.heatmap(cm_normd, **heatmap_kwargs)\n",
        "    \n",
        "    plt.title(f\"{model.__class__.__name__} Classification\")\n",
        "    plt.ylabel(\"Ground-truth labels\")\n",
        "    plt.xlabel(\"Predicted labels\")\n",
        "    plt.plot()"
      ],
      "execution_count": 67,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fK7_ZX4FrlU6"
      },
      "source": [
        "def train_and_grade(tfidf, model, train_data, train_labels, test_data, test_labels):\n",
        "    preds = train(tfidf, model, train_data, train_labels, test_data)\n",
        "    grade(model, preds, test_labels)"
      ],
      "execution_count": 68,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 458
        },
        "id": "RoagZO1xrlU7",
        "outputId": "4664e3d5-aa09-4abe-a587-4760525d4d75"
      },
      "source": [
        "svm_lsvc = LinearSVC(class_weight=\"balanced\")\n",
        "clf = CalibratedClassifierCV(svm_lsvc)\n",
        "train_and_grade(tfidf, clf, X_train, y_train, X_valid, y_valid)"
      ],
      "execution_count": 69,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "    Negative       0.81      0.80      0.81    153794\n",
            "    Positive       0.80      0.82      0.81    153706\n",
            "\n",
            "    accuracy                           0.81    307500\n",
            "   macro avg       0.81      0.81      0.81    307500\n",
            "weighted avg       0.81      0.81      0.81    307500\n",
            "\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xNs61s3IdHBG"
      },
      "source": [
        "import pickle\r\n",
        "pickle.dump(clf, open( \"save.p\", \"wb\" ) )"
      ],
      "execution_count": 71,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Iwzlex2RrlU7"
      },
      "source": [
        "preds = clf.predict(tfidf.transform(X_test))\n",
        "preds_proba = clf.predict_proba(tfidf.transform(X_test))"
      ],
      "execution_count": 72,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "cP_fEQrQgeZt",
        "outputId": "1cad772c-9f93-4587-9eb9-c03e962149ca"
      },
      "source": [
        "preds"
      ],
      "execution_count": 78,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array(['Negative', 'Negative', 'Negative', ..., 'Negative', 'Negative',\n",
              "       'Positive'], dtype='<U8')"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 78
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 337
        },
        "id": "6HABFj97rlU8",
        "outputId": "896bc342-fa19-4544-9060-ac98ff5a875a"
      },
      "source": [
        "plt.hist(preds)"
      ],
      "execution_count": 73,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(array([502239.,      0.,      0.,      0.,      0.,      0.,      0.,\n",
              "             0.,      0., 522761.]),\n",
              " array([0. , 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1. ]),\n",
              " <a list of 10 Patch objects>)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 73
        },
        {
          "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": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "fJbDO2xdhAVf",
        "outputId": "a52d7b4c-1630-45ba-8c0b-1340bf6204a7"
      },
      "source": [
        "pred_neut"
      ],
      "execution_count": 87,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array(['Negative', 'Negative', 'Negative', ..., 'Neutral', 'Negative',\n",
              "       'Positive'], dtype='<U8')"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 87
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 341
        },
        "id": "N9WNif3t5v9Y",
        "outputId": "56f4f8c3-1edc-4c3c-cf66-ecf57c5dd25b"
      },
      "source": [
        "f = lambda x: 'Negative' if x>0.6 else ('Neutral' if x>0.4 else 'Positive')\r\n",
        "pred_neut = np.array([f(xi) for xi in preds_proba[:,0]])\r\n",
        "plt.hist(pred_neut)"
      ],
      "execution_count": 174,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(array([436673.,      0.,      0.,      0.,      0., 454451.,      0.,\n",
              "             0.,      0., 133876.]),\n",
              " array([0. , 0.2, 0.4, 0.6, 0.8, 1. , 1.2, 1.4, 1.6, 1.8, 2. ]),\n",
              " <a list of 10 Patch objects>)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 174
        },
        {
          "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": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "B-SU3U2rvNlc",
        "outputId": "17d6fa1a-b6e1-4701-ae2a-549bbae4e680"
      },
      "source": [
        "!pip install fasttext"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Collecting fasttext\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/f8/85/e2b368ab6d3528827b147fdb814f8189acc981a4bc2f99ab894650e05c40/fasttext-0.9.2.tar.gz (68kB)\n",
            "\r\u001b[K     |████▊                           | 10kB 17.8MB/s eta 0:00:01\r\u001b[K     |█████████▌                      | 20kB 21.3MB/s eta 0:00:01\r\u001b[K     |██████████████▎                 | 30kB 10.1MB/s eta 0:00:01\r\u001b[K     |███████████████████             | 40kB 8.1MB/s eta 0:00:01\r\u001b[K     |███████████████████████▉        | 51kB 4.4MB/s eta 0:00:01\r\u001b[K     |████████████████████████████▋   | 61kB 4.5MB/s eta 0:00:01\r\u001b[K     |████████████████████████████████| 71kB 3.3MB/s \n",
            "\u001b[?25hRequirement already satisfied: pybind11>=2.2 in /usr/local/lib/python3.6/dist-packages (from fasttext) (2.6.1)\n",
            "Requirement already satisfied: setuptools>=0.7.0 in /usr/local/lib/python3.6/dist-packages (from fasttext) (50.3.2)\n",
            "Requirement already satisfied: numpy in /usr/local/lib/python3.6/dist-packages (from fasttext) (1.19.4)\n",
            "Building wheels for collected packages: fasttext\n",
            "  Building wheel for fasttext (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "  Created wheel for fasttext: filename=fasttext-0.9.2-cp36-cp36m-linux_x86_64.whl size=3039080 sha256=d2f92a2434dab33370b44bc8adf6e51b772d25cebbe965b8e2189577d40ee024\n",
            "  Stored in directory: /root/.cache/pip/wheels/98/ba/7f/b154944a1cf5a8cee91c154b75231136cc3a3321ab0e30f592\n",
            "Successfully built fasttext\n",
            "Installing collected packages: fasttext\n",
            "Successfully installed fasttext-0.9.2\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "UlAC7POYrlU8"
      },
      "source": [
        "import fasttext"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DG9tEAx3XjZS"
      },
      "source": [
        "def shuffle(series):\r\n",
        "  series.index = np.linspace(0, len(series)-1, len(series))\r\n",
        "  series = series.iloc[np.random.permutation(len(pos_))]\r\n",
        "  series.index = np.linspace(0, len(series)-1, len(series))\r\n",
        "  return series"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ECpkSIPPDvp3"
      },
      "source": [
        "test_percentage = 0.1\n",
        "validate_percentage = 0.1\n",
        "pos = shuffle(X_all[y_all==4])\n",
        "neg = shuffle(X_all[y_all==0])\n",
        "\n",
        "num_neg_test = int(test_percentage * len(neg) )\n",
        "num_neg_validate = int(test_percentage * len(neg) )\n",
        "\n",
        "num_pos_test = int(test_percentage * len(pos) )\n",
        "num_pos_validate = int(validate_percentage * len(pos) )"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "wXniOYvsrlU9",
        "outputId": "80de45e0-1759-4b59-c2a7-eb2ef62132e2"
      },
      "source": [
        "try:\n",
        "    fileTrain = open(\"train.txt\",\"w\",errors=\"namereplace\")\n",
        "    fileValid = open(\"valid.txt\",\"w\",errors=\"namereplace\")\n",
        "    fileTest = open(\"test.txt\",\"w\",errors=\"namereplace\")\n",
        "    print(\"pos \",len(pos),\"neg\",len(neg))\n",
        "    i = 0\n",
        "    for neg_line in neg:\n",
        "        # if i % 1e5==0: print(neg_line,'\\n') \n",
        "        if (i < num_neg_test):\n",
        "            fileTest.write(\"__label__0 \"+(neg_line)+\"\\n\")\n",
        "        elif (i < num_neg_test + num_neg_validate):\n",
        "            fileValid.write(\"__label__0 \"+(neg_line)+\"\\n\")\n",
        "        else:\n",
        "            fileTrain.write(\"__label__0 \"+(neg_line)+\"\\n\")\n",
        "        i +=1\n",
        "    i = 0 \n",
        "    for pos_line in pos:\n",
        "        # if i % 1e5==0: print(pos_line,'\\n') \n",
        "        if(i < num_neg_test):\n",
        "            fileTest.write(\"__label__4 \"+(pos_line)+\"\\n\")\n",
        "        elif (i < num_neg_test + num_neg_validate):\n",
        "            fileValid.write(\"__label__4 \"+(pos_line)+\"\\n\")\n",
        "        else:\n",
        "            fileTrain.write(\"__label__4 \"+(pos_line)+\"\\n\")\n",
        "        i +=1\n",
        "# except:\n",
        "#     print(\"Error try again\",flush = True)\n",
        "finally:\n",
        "    fileTrain.close()\n",
        "    fileValid.close()\n",
        "    fileTest.close()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "pos  2050000 neg 2050000\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PTH571jYrlU-"
      },
      "source": [
        "model = fasttext.train_supervised(input = \"ready_for_fasttext.txt\",autotuneValidationFile = \"valid_fastText.txt\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "dhwQrDDa0d32"
      },
      "source": [
        "model_no_val = fasttext.train_supervised(input = \"train.txt\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tS0Nl5lP08Rf"
      },
      "source": [
        "hyper_params = {\"lr\": 0.01,\r\n",
        "    \"epoch\": 70,\r\n",
        "    \"wordNgrams\": 2,\r\n",
        "    \"dim\": 20}"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1zH2HSAf09Gh"
      },
      "source": [
        "model_hyper = fasttext.train_supervised(input=\"train.txt\", **hyper_params)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "txBG1azpCiA-",
        "outputId": "fccef495-9870-4a92-eb9b-bcdf3a6a1283"
      },
      "source": [
        "model.test(\"test.txt\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(410000, 0.5007780487804878, 0.5007780487804878)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 124
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "pTrva4nCHQeV",
        "outputId": "6fac5af3-50ac-4c2d-8ebb-151d1033b852"
      },
      "source": [
        "model.test(\"train.txt\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(3280000, 0.5041326219512196, 0.5041326219512196)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 84
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "F56MY0ptHQns",
        "outputId": "d5ec2ed8-4aff-4d6f-ed85-c8f30d966a43"
      },
      "source": [
        "model.test(\"valid.txt\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(410000, 0.503980487804878, 0.503980487804878)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 85
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qWLh6groTAvy"
      },
      "source": [
        "def test_model(model,test_set):\r\n",
        "  correct = 0\r\n",
        "  for i in range(len(test_set)):\r\n",
        "    if model.predict(test_set['text'][i])[0][0] == '__label__4':\r\n",
        "      temp = 4\r\n",
        "    else:\r\n",
        "      temp = 0\r\n",
        "    if temp == test_set['target'][i]:\r\n",
        "      correct = correct + 1\r\n",
        "\r\n",
        "  accuracy = correct/len(test_set)\r\n",
        "  return accuracy"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-ALBwz-z1dLe"
      },
      "source": [
        "test_model(model)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "5ik5Ws4XxUR2",
        "outputId": "ad082a46-6b08-4f00-8483-a339fba5b935"
      },
      "source": [
        "test_model(model_no_val)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.5059853366568944"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 221
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "rhgxJLz21eAm",
        "outputId": "4e8a4096-b1f9-4ee0-f2c0-30d4c1a47607"
      },
      "source": [
        "test_model(model_hyper)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.5893873688421032"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 238
        }
      ]
    }
  ]
}