Sentimental_Analysis_pro.ipynb

{
  "cells": [
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "BK1WNcCn1XRz",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "#from google.colab import drive\n",
        "#drive.mount('/content/drive')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "x6xrRjpk8rAU",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "#!unzip '/content/drive/My Drive/arabic-sentiment-analysis-dataset-ss2030-dataset.zip' -d /content/"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "nQWlxoCrEmWt",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "# Import Libraries"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 63
        },
        "id": "mh23RjkB957t",
        "outputId": "431927e8-6921-4929-942b-1594a04d6731",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "data": {
            "text/html": [
              "<p style=\"color: red;\">\n",
              "The default version of TensorFlow in Colab will soon switch to TensorFlow 2.x.<br>\n",
              "We recommend you <a href=\"https://www.tensorflow.org/guide/migrate\" target=\"_blank\">upgrade</a> now \n",
              "or ensure your notebook will continue to use TensorFlow 1.x via the <code>%tensorflow_version 1.x</code> magic:\n",
              "<a href=\"https://colab.research.google.com/notebooks/tensorflow_version.ipynb\" target=\"_blank\">more info</a>.</p>\n"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "import pandas as pd \n",
        "import string \n",
        "import nltk\n",
        "from nltk.stem.isri import ISRIStemmer\n",
        "import re\n",
        "from sklearn.model_selection import train_test_split\n",
        "import tensorflow as tf\n",
        "import numpy as np\n",
        "from tensorflow.python.keras.preprocessing import sequence\n",
        "from tensorflow.python.keras.preprocessing import text\n",
        "from tensorflow.python.keras import models"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ST1vdYLB6yCF",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "# Read DataSet"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 204
        },
        "id": "5DGv-Nka957y",
        "outputId": "caa0c7a3-1f74-44b5-a2f2-b37ae525131d",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "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>tweets</th>\n",
              "      <th>labels</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>امتلأت هذه الفترة بالكثير من الجراح والطعنات ا...</td>\n",
              "      <td>Neg</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>سامي الجابر غبي اثبت نفسك انك ممتاز وانك كفاءة...</td>\n",
              "      <td>Neg</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>لاتستطيع تقييم العمل بشكل مطلق ونهائي . . . لو...</td>\n",
              "      <td>Neg</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>تبغى تتابعه بكيفك هذا رايك انا المسلسل ماعجبني...</td>\n",
              "      <td>Neg</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>التشقير بالليزر ضعيف ولايخفف الشعر ولايوجد حل ...</td>\n",
              "      <td>Neg</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                                              tweets labels\n",
              "0  امتلأت هذه الفترة بالكثير من الجراح والطعنات ا...    Neg\n",
              "1  سامي الجابر غبي اثبت نفسك انك ممتاز وانك كفاءة...    Neg\n",
              "2  لاتستطيع تقييم العمل بشكل مطلق ونهائي . . . لو...    Neg\n",
              "3  تبغى تتابعه بكيفك هذا رايك انا المسلسل ماعجبني...    Neg\n",
              "4  التشقير بالليزر ضعيف ولايخفف الشعر ولايوجد حل ...    Neg"
            ]
          },
          "execution_count": 7,
          "metadata": {
            "tags": []
          },
          "output_type": "execute_result"
        }
      ],
      "source": [
        "Neg_Tweets = pd.read_csv('/content/NegTrain.txt', encoding='windows-1256',names=[ \"tweets\"]) # dataframe \n",
        "Neg_Tweets['labels'] = 'Neg' #assign labels to every row\n",
        "Neg_Tweets.head()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 204
        },
        "id": "nX-n7Jko9573",
        "outputId": "abd310c3-0c04-434a-ff45-e7e02b2ed821",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "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>tweets</th>\n",
              "      <th>labels</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>لا تجد قوما أشد سعادة في الدنيا ولا أوسع صدورا...</td>\n",
              "      <td>Pos</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>! رسايلك صوتك سوالفك طاريك والله إنها للقلب مص...</td>\n",
              "      <td>Pos</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>سعادة الآخرين لن تؤخذ من سعادتك</td>\n",
              "      <td>Pos</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>المعصية لها توبة</td>\n",
              "      <td>Pos</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>مساء السعادة</td>\n",
              "      <td>Pos</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                                              tweets labels\n",
              "0  لا تجد قوما أشد سعادة في الدنيا ولا أوسع صدورا...    Pos\n",
              "1  ! رسايلك صوتك سوالفك طاريك والله إنها للقلب مص...    Pos\n",
              "2                   سعادة الآخرين لن تؤخذ من سعادتك     Pos\n",
              "3                                  المعصية لها توبة     Pos\n",
              "4                                      مساء السعادة     Pos"
            ]
          },
          "execution_count": 8,
          "metadata": {
            "tags": []
          },
          "output_type": "execute_result"
        }
      ],
      "source": [
        "Pos_Tweets = pd.read_csv('/content/PosTrain.txt', encoding='windows-1256',names=[ \"tweets\"])\n",
        "Pos_Tweets['labels'] = 'Pos'\n",
        "Pos_Tweets.head()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "qOGST90ptJtb",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "Test_Pos_Tweets = pd.read_csv('/content/PosTest.txt', encoding='windows-1256',names=[ \"tweets\"])\n",
        "Test_Pos_Tweets['labels'] = 'Pos'"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "p-41SROem8Wf",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "Test_Neg_Tweets = pd.read_csv('/content/NegTest.txt', encoding='windows-1256',names=[ \"tweets\"])\n",
        "Test_Neg_Tweets['labels'] = 'Neg'"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 204
        },
        "id": "X50p7gmD-4rd",
        "outputId": "57935849-6e82-4c7e-dd57-d44744a25e20",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "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>tweets</th>\n",
              "      <th>labels</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>حقوق المرأة 💚💚💚 https://t.co/Mzf90Ta5g1</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>RT @___IHAVENOIDEA: حقوق المرأة في الإسلام. ht...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>RT @saud_talep: Retweeted لجنة التنمية بشبرا (...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>RT @MojKsa: حقوق المرأة التي تضمنها لها وزارة ...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>RT @abm112211: ولي امر الزوجة او ولي الزوجة او...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                                              tweets  labels\n",
              "0            حقوق المرأة 💚💚💚 https://t.co/Mzf90Ta5g1       1\n",
              "1  RT @___IHAVENOIDEA: حقوق المرأة في الإسلام. ht...       1\n",
              "2  RT @saud_talep: Retweeted لجنة التنمية بشبرا (...       1\n",
              "3  RT @MojKsa: حقوق المرأة التي تضمنها لها وزارة ...       1\n",
              "4  RT @abm112211: ولي امر الزوجة او ولي الزوجة او...       1"
            ]
          },
          "execution_count": 11,
          "metadata": {
            "tags": []
          },
          "output_type": "execute_result"
        }
      ],
      "source": [
        "new_data=pd.read_csv('/content/Arabic Sentiment Analysis Dataset - SS2030.csv')\n",
        "new_data.columns=['tweets','labels']\n",
        "new_data.head()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "sIa_oNXlUCOM",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "new_data.labels=new_data.labels.astype('int64')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "TRGn2irZUSko",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "new_data['labels']= new_data['labels'].map( {1:'Pos', 0:'Neg'} )"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0il_UcvztOal",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "# Combine All Test And Train Set"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "lw5vhOQS9577",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "tweets=pd.concat([Pos_Tweets,Neg_Tweets,Test_Neg_Tweets,Test_Pos_Tweets,new_data],axis=0, ignore_index=True,sort=False )"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lPAJyqqaGfoL",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "# Exploratory Data Analysis\n",
        "Let's check out some of the stats with some plots and the built-in methods in pandas!\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 173
        },
        "id": "QVFggMuB957-",
        "outputId": "2e374b64-8f9d-4086-8ac0-8a138d7989f7",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "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>tweets</th>\n",
              "      <th>labels</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>count</th>\n",
              "      <td>15359</td>\n",
              "      <td>15364</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>unique</th>\n",
              "      <td>15249</td>\n",
              "      <td>2</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>top</th>\n",
              "      <td>ممتاز</td>\n",
              "      <td>Neg</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>freq</th>\n",
              "      <td>11</td>\n",
              "      <td>7971</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "        tweets labels\n",
              "count    15359  15364\n",
              "unique   15249      2\n",
              "top     ممتاز     Neg\n",
              "freq        11   7971"
            ]
          },
          "execution_count": 15,
          "metadata": {
            "tags": []
          },
          "output_type": "execute_result"
        }
      ],
      "source": [
        "tweets.describe()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "U6IBkEBOCJ3M",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "Let's use groupby to use describe by label, this way we can begin to think about the features that separate Pos and Neg!"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 173
        },
        "id": "69rrtqkt958A",
        "outputId": "fa218d76-17de-4d90-b9a1-c7b06d480ace",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "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 tr th {\n",
              "        text-align: left;\n",
              "    }\n",
              "\n",
              "    .dataframe thead tr:last-of-type th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr>\n",
              "      <th></th>\n",
              "      <th colspan=\"4\" halign=\"left\">tweets</th>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th></th>\n",
              "      <th>count</th>\n",
              "      <th>unique</th>\n",
              "      <th>top</th>\n",
              "      <th>freq</th>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>labels</th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "      <th></th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>Neg</th>\n",
              "      <td>7969</td>\n",
              "      <td>7923</td>\n",
              "      <td>للأسف</td>\n",
              "      <td>8</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>Pos</th>\n",
              "      <td>7390</td>\n",
              "      <td>7331</td>\n",
              "      <td>ممتاز</td>\n",
              "      <td>11</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "       tweets                    \n",
              "        count unique     top freq\n",
              "labels                           \n",
              "Neg      7969   7923  للأسف     8\n",
              "Pos      7390   7331  ممتاز    11"
            ]
          },
          "execution_count": 16,
          "metadata": {
            "tags": []
          },
          "output_type": "execute_result"
        }
      ],
      "source": [
        "tweets.groupby('labels').describe()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dRflE3t3Jl7J",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "# Text Pre-processing"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "NovNJ9MCVjUM",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "### Check Null Values"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 68
        },
        "id": "bjwc8lcXVnA1",
        "outputId": "7d7e575b-5e95-49cf-b51d-fa8874616da5",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "tweets    5\n",
              "labels    0\n",
              "dtype: int64"
            ]
          },
          "execution_count": 17,
          "metadata": {
            "tags": []
          },
          "output_type": "execute_result"
        }
      ],
      "source": [
        "tweets.isnull().sum()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 68
        },
        "id": "FLp2CazAVc5e",
        "outputId": "23dd983f-24a0-4106-a63f-e4453370db85",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "tweets    0\n",
              "labels    0\n",
              "dtype: int64"
            ]
          },
          "execution_count": 18,
          "metadata": {
            "tags": []
          },
          "output_type": "execute_result"
        }
      ],
      "source": [
        "tweets.dropna(inplace=True) #drop null rows\n",
        "tweets.isnull().sum()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Sgc-IY75VxNp",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "### Check Duplicate Recode"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "id": "abcuGh5mVbjy",
        "outputId": "ad9c44a0-ef4b-468e-da67-cfcdc2df63c7",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "110"
            ]
          },
          "execution_count": 19,
          "metadata": {
            "tags": []
          },
          "output_type": "execute_result"
        }
      ],
      "source": [
        "tweets.tweets.duplicated().sum()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 419
        },
        "id": "FzZDrRIw-cSr",
        "outputId": "ea83c154-b5cc-48da-bcaa-40152485c092",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "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>tweets</th>\n",
              "      <th>labels</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>283</th>\n",
              "      <td>مساء الخير</td>\n",
              "      <td>Pos</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>345</th>\n",
              "      <td>مساء السعادة</td>\n",
              "      <td>Pos</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>615</th>\n",
              "      <td>رائع</td>\n",
              "      <td>Pos</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>822</th>\n",
              "      <td>ممتاز</td>\n",
              "      <td>Pos</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>843</th>\n",
              "      <td>ممتاز</td>\n",
              "      <td>Pos</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>13548</th>\n",
              "      <td>@seen_180 انا قدها وانا بصراحه مع تعدد الزوجات</td>\n",
              "      <td>Pos</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>13560</th>\n",
              "      <td>RT @bbahth2: #تعدد_الزوجات سبب لنشر العفاف وإع...</td>\n",
              "      <td>Pos</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>13644</th>\n",
              "      <td>@Akhbaar24 حسبي الله ونعم الوكيل عليهم ودعمهم ...</td>\n",
              "      <td>Neg</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>13647</th>\n",
              "      <td>RT @SAUD_SAIFALNASR: #ابشركم_حنا_بخير تكاليف ا...</td>\n",
              "      <td>Neg</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>13648</th>\n",
              "      <td>الراتب موظف لا يكفي لمدة 15 يوم \\n #حملة_تعديل...</td>\n",
              "      <td>Neg</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>110 rows × 2 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "                                                  tweets labels\n",
              "283                                          مساء الخير     Pos\n",
              "345                                        مساء السعادة     Pos\n",
              "615                                                رائع     Pos\n",
              "822                                               ممتاز     Pos\n",
              "843                                               ممتاز     Pos\n",
              "...                                                  ...    ...\n",
              "13548     @seen_180 انا قدها وانا بصراحه مع تعدد الزوجات    Pos\n",
              "13560  RT @bbahth2: #تعدد_الزوجات سبب لنشر العفاف وإع...    Pos\n",
              "13644  @Akhbaar24 حسبي الله ونعم الوكيل عليهم ودعمهم ...    Neg\n",
              "13647  RT @SAUD_SAIFALNASR: #ابشركم_حنا_بخير تكاليف ا...    Neg\n",
              "13648  الراتب موظف لا يكفي لمدة 15 يوم \\n #حملة_تعديل...    Neg\n",
              "\n",
              "[110 rows x 2 columns]"
            ]
          },
          "execution_count": 20,
          "metadata": {
            "tags": []
          },
          "output_type": "execute_result"
        }
      ],
      "source": [
        "tweets.loc[tweets.tweets.duplicated(),:] #duplicate recode"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "6YZEfSn__QVS",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "tweets.drop_duplicates(subset=['tweets'],keep='first',inplace=True) # drop duplicates and kepp one of them"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "id": "8aLlXUxh_j3A",
        "outputId": "55834976-9b67-480e-900f-179295783e5e",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "0"
            ]
          },
          "execution_count": 22,
          "metadata": {
            "tags": []
          },
          "output_type": "execute_result"
        }
      ],
      "source": [
        "tweets.tweets.duplicated().sum()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0swUpsO4_Q8D",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "## Clean Text\n",
        "Let's create a function that will process the string in the tweets column, then we can just use apply() in pandas do process all the text in the DataFrame."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "jIbgoYVHJhm6",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "# Arabic diacritics\n",
        "arabic_punct = ''' ` ÷ × ؛ < > _ ( ) * & ^ % ] [ ـ ، / : \" ? ؟ . , ' { } ~ ¦ + | !  ”  …  “ –   ـ  '''\n",
        "arabic_diacritics = '''  َ     ُ       ِ      ّ       ً      ٌ       ٍ      ْ     '''\n",
        "english_punctuations = string.punctuation\n",
        "\n",
        "arabic_punct = arabic_punct.split()\n",
        "arabic_diacritics = arabic_diacritics.split()\n",
        "\n",
        "# Arabic punctuations and dicritis \n",
        "punctuations = str(arabic_punct  + arabic_diacritics)\n",
        "punctuations=punctuations+english_punctuations\n",
        "# Arabic stopwords\n",
        "asw = open('/content//stop_words.txt', encoding=\"utf8\").read()\n",
        "\n",
        "arabicStopWords =  set(asw.split())"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "zIL7NmBK958F",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "# remove arabic diacritics\n",
        "def remove_diacritics(text):\n",
        "\tarstemmer = ISRIStemmer()\n",
        "\tresult = arstemmer.norm(text, num=1) #  remove diacritics which representing Arabic short vowels\n",
        "\treturn result"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "zzjnLXaO958I",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "regex = re.compile('[%s]' % re.escape(punctuations))\n",
        "# remove punctcutions\n",
        "def remove_punct(word):\n",
        "  return regex.sub(' ', word)\n",
        "  #return word.translate(str.maketrans('','',punctuations))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "_8NuwMBC9Rbk",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "def reove_eng_word(word):\n",
        "    text=re.sub(r'[0-9]', '' , word)\n",
        "    return re.sub(r'[a-z]+', '' , text)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "B62bH35GBj4I",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "emoji_pattern = re.compile(\"[\"\n",
        "                           u\"\\U0001F600-\\U0001F64F\"  # emoticons\n",
        "                           u\"\\U0001F300-\\U0001F5FF\"  # symbols & pictographs\n",
        "                           u\"\\U0001F680-\\U0001F6FF\"  # transport & map symbols\n",
        "                           u\"\\U0001F1E0-\\U0001F1FF\"  # flags (iOS)\n",
        "                           u\"\\U00002702-\\U000027B0\"\n",
        "                           u\"\\U000024C2-\\U0001F251\"\n",
        "                           \"]+\", flags=re.UNICODE)\n",
        "def remove_emojies(string):\n",
        "    return emoji_pattern.sub(r'', string)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "YvMKzWGH958K",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "def process_text(text, removePunct=True, removeSW=False, removeNum=True,removeEng=True,remove_emoji=True):\n",
        "\t'''\n",
        "\t1. remove arabic diacritics\n",
        "\t2. remove punctcutions\n",
        "\t3. remove stopwords if removeSW=True\n",
        "\t4. remove numbers if removeNum=True\n",
        "\t'''\n",
        "\ttext = remove_diacritics(text)  # remove arabic diacritics\n",
        "\tword_list = nltk.tokenize.wordpunct_tokenize(text.lower())\n",
        "\tif removePunct:\n",
        "\t\tword_list = [ w for w in word_list if not w in punctuations ]\n",
        "\t\tword_list = [ remove_punct(w) for w in word_list ]\n",
        "\n",
        "\tif removeSW: word_list = [ w for w in word_list if not w in arabicStopWords ]\n",
        "\tif removeEng: \n",
        "\t\tword_list  = [ reove_eng_word(w) for w in word_list ]\n",
        "\tif remove_emoji:\n",
        "\t\tword_list  = [ remove_emojies(w) for w in word_list ]\n",
        "\t#if removeNum: word_list = [ w for w in word_list if not w.isdigit() ]\n",
        "\tword_list = [ w for w in word_list if w] # remove empty words\n",
        "\t#return ' '.join(word_list)\n",
        "\treturn (re.sub(' +', ' ', ' '.join(word_list))).strip() # remove spaces more then 1 ."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "id": "EQso5ox4958Q",
        "outputId": "28506b1c-fcd8-4bc0-9fc1-fca3e48263cc",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "'! رسايلك صوتك سوالفك طاريك والله إنها للقلب مصدر سعادة ء ء'"
            ]
          },
          "execution_count": 29,
          "metadata": {
            "tags": []
          },
          "output_type": "execute_result"
        }
      ],
      "source": [
        "tweets['tweets'][1]"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "id": "KaFlkOV4958T",
        "outputId": "d3a27a15-351d-423e-a880-667e052fb3db",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "'رسايلك صوتك سوالفك طاريك والله إنها للقلب مصدر سعادة ء ء'"
            ]
          },
          "execution_count": 30,
          "metadata": {
            "tags": []
          },
          "output_type": "execute_result"
        }
      ],
      "source": [
        "process_text(tweets['tweets'][1])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "id": "jb-besbmWFmL",
        "outputId": "4fb1e075-70a6-4b76-86ea-593f928441df",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "'RT @___IHAVENOIDEA: حقوق المرأة في الإسلام. https://t.co/ps3qNw1CbB'"
            ]
          },
          "execution_count": 31,
          "metadata": {
            "tags": []
          },
          "output_type": "execute_result"
        }
      ],
      "source": [
        "new_data['tweets'][1]"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "id": "PjNHYhstWKAH",
        "outputId": "516303e8-26ec-4481-999a-16afd609c2f9",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "'حقوق المرأة في الإسلام'"
            ]
          },
          "execution_count": 32,
          "metadata": {
            "tags": []
          },
          "output_type": "execute_result"
        }
      ],
      "source": [
        "process_text(new_data['tweets'][1])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "id": "TgcxrYGXXBWj",
        "outputId": "1c4b4748-2468-4340-e204-6e399813bd8a",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "'حقوق المرأة 💚💚💚 https://t.co/Mzf90Ta5g1'"
            ]
          },
          "execution_count": 33,
          "metadata": {
            "tags": []
          },
          "output_type": "execute_result"
        }
      ],
      "source": [
        "new_data['tweets'][0]"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "id": "N3X7TZQ1XFQ1",
        "outputId": "bccac4a1-a6bd-4d60-fedd-d7aebccb44c8",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "'حقوق المرأة'"
            ]
          },
          "execution_count": 34,
          "metadata": {
            "tags": []
          },
          "output_type": "execute_result"
        }
      ],
      "source": [
        "process_text(new_data['tweets'][0])"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9fUeaQ4UKwc1",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "Apply this function now to all tweets"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "VPKo_6_U958b",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "tweets['tweets'] = tweets['tweets'].apply(process_text) # apply it to all tweets"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 204
        },
        "id": "jPhhetA0NRUA",
        "outputId": "681f3e25-a515-49da-a045-f3a5494808e9",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "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>tweets</th>\n",
              "      <th>labels</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>لا تجد قوما أشد سعادة في الدنيا ولا أوسع صدورا...</td>\n",
              "      <td>Pos</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>رسايلك صوتك سوالفك طاريك والله إنها للقلب مصدر...</td>\n",
              "      <td>Pos</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>سعادة الآخرين لن تؤخذ من سعادتك</td>\n",
              "      <td>Pos</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>المعصية لها توبة</td>\n",
              "      <td>Pos</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>مساء السعادة</td>\n",
              "      <td>Pos</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                                              tweets labels\n",
              "0  لا تجد قوما أشد سعادة في الدنيا ولا أوسع صدورا...    Pos\n",
              "1  رسايلك صوتك سوالفك طاريك والله إنها للقلب مصدر...    Pos\n",
              "2                    سعادة الآخرين لن تؤخذ من سعادتك    Pos\n",
              "3                                   المعصية لها توبة    Pos\n",
              "4                                       مساء السعادة    Pos"
            ]
          },
          "execution_count": 36,
          "metadata": {
            "tags": []
          },
          "output_type": "execute_result"
        }
      ],
      "source": [
        "tweets.head(5)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "2qOIaQ-lCmqx",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "As we continue our analysis we want to start thinking about the features we are going to be using. This goes along with the general idea of [feature engineering](https://en.wikipedia.org/wiki/Feature_engineering). The better your domain knowledge on the data, the better your ability to engineer more features from it. I encourage you to read up on the topic!\n",
        "\n",
        "Let's make a new column to detect how long the text messages are:"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 204
        },
        "id": "iKxQSBx1WXcN",
        "outputId": "84edf362-e16d-4994-cbe3-6a2551628b1d",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "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>tweets</th>\n",
              "      <th>labels</th>\n",
              "      <th>length</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>لا تجد قوما أشد سعادة في الدنيا ولا أوسع صدورا...</td>\n",
              "      <td>Pos</td>\n",
              "      <td>131</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>رسايلك صوتك سوالفك طاريك والله إنها للقلب مصدر...</td>\n",
              "      <td>Pos</td>\n",
              "      <td>56</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>سعادة الآخرين لن تؤخذ من سعادتك</td>\n",
              "      <td>Pos</td>\n",
              "      <td>31</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>المعصية لها توبة</td>\n",
              "      <td>Pos</td>\n",
              "      <td>16</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>مساء السعادة</td>\n",
              "      <td>Pos</td>\n",
              "      <td>12</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                                              tweets labels  length\n",
              "0  لا تجد قوما أشد سعادة في الدنيا ولا أوسع صدورا...    Pos     131\n",
              "1  رسايلك صوتك سوالفك طاريك والله إنها للقلب مصدر...    Pos      56\n",
              "2                    سعادة الآخرين لن تؤخذ من سعادتك    Pos      31\n",
              "3                                   المعصية لها توبة    Pos      16\n",
              "4                                       مساء السعادة    Pos      12"
            ]
          },
          "execution_count": 37,
          "metadata": {
            "tags": []
          },
          "output_type": "execute_result"
        }
      ],
      "source": [
        "tweets['length'] = tweets['tweets'].str.len() #check Leanth of Each tweet\n",
        "tweets.head()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 170
        },
        "id": "IXXx5QdoWKNb",
        "outputId": "ac557a37-f1e9-4a1f-e6fc-4f6f560f68ec",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "count    15249.000000\n",
              "mean        87.172601\n",
              "std         59.223386\n",
              "min          0.000000\n",
              "25%         42.000000\n",
              "50%         76.000000\n",
              "75%        118.000000\n",
              "max        280.000000\n",
              "Name: length, dtype: float64"
            ]
          },
          "execution_count": 38,
          "metadata": {
            "tags": []
          },
          "output_type": "execute_result"
        }
      ],
      "source": [
        "tweets.length.describe()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "tbarS-mhDH43",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "Maximum tweet leanth is 185."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "iR6TU7hs959a",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "tweets = tweets.sample(frac=1,random_state=30).reset_index(drop=True) #shuffle data frame"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "86cw7SzmDBQf",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "# Data Visualization\n",
        "Let's visualize this! Let's do the imports:"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "phsgrv6M-ZJd",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "import matplotlib.pyplot as plt\n",
        "import seaborn as sns\n",
        "\n",
        "%matplotlib inline\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 296
        },
        "id": "L0y51cWQDWCX",
        "outputId": "0a15fd8d-2f5a-4240-e2e8-9aea650d6163",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7f9998fd8ef0>"
            ]
          },
          "execution_count": 41,
          "metadata": {
            "tags": []
          },
          "output_type": "execute_result"
        },
        {
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYAAAAEGCAYAAABsLkJ6AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nhBBilJIEIIQQo5QkACGEGKX+P5f8/Rh5YNmJAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "sns.set_style('whitegrid')\n",
        "sns.distplot(tweets['length'],bins=25)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 340
        },
        "id": "zVo9-Yf2D_tw",
        "outputId": "8a816a75-9b6c-4fdd-85a5-2fc29ceb347b",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "array([<matplotlib.axes._subplots.AxesSubplot object at 0x7f9998818438>,\n",
              "       <matplotlib.axes._subplots.AxesSubplot object at 0x7f99987ca198>],\n",
              "      dtype=object)"
            ]
          },
          "execution_count": 42,
          "metadata": {
            "tags": []
          },
          "output_type": "execute_result"
        },
        {
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAtsAAAEQCAYAAABht7hCAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 864x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "tweets.hist(column='length', by='labels', bins=25,figsize=(12,4))"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "2-Xh0mlmdizE",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "# Prepare Data"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "I7bKUtY5bO3l",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "\n",
        "from tensorflow.python.keras import initializers\n",
        "from tensorflow.python.keras import regularizers\n",
        "\n",
        "from tensorflow.python.keras.layers import Dense\n",
        "from tensorflow.python.keras.layers import Dropout\n",
        "from tensorflow.python.keras.layers import Embedding\n",
        "from tensorflow.python.keras.layers import MaxPooling1D\n",
        "from tensorflow.python.keras.layers import GlobalMaxPool1D"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "qMBe_iCDMMso",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "### Encoding\n",
        "Encode class data to numeric format"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "GBhAYLK3esFN",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "tweets['labels']= tweets['labels'].map( {'Pos':1, 'Neg': 0} ).astype(int)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "US8h0NNQMBM0",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "# Train_Test_Split"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "gTUx-jq-LeP0",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "We will convert our data in three parts, training,validation and testing. We train our model on traing and tune on validation and then test it on testing data"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "vb8XpgFwdxJl",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "X_train, X_test, y_train, y_test = train_test_split(\n",
        "    tweets['tweets'], tweets['labels'], random_state=0,stratify=tweets['labels'],test_size=0.30,shuffle=True)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "7RgDTGZmslsy",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "X_train, X_val, y_train, y_val = train_test_split(\n",
        "    X_train, y_train, random_state=0,stratify=y_train,test_size=0.10,shuffle=True)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "knYRmoZxNTFl",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "Before our data can be fed to a model, it needs to be transformed to a format the model can understand. This means that we will need to convert the texts into numerical vectors."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "RYbKnrBg0Q3k",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "# Vectorization parameters\n",
        "# Limit on the number of features. We use the top 20K features.\n",
        "TOP_K = 20000\n",
        "# Limit on the length of text sequences. Sequences longer than this will be truncated.\n",
        "MAX_SEQUENCE_LENGTH = 200\n",
        "\n",
        "def sequence_vectorize(train_texts, val_texts):\n",
        "    \"\"\"\n",
        "    Vectorizes texts as sequence vectors.\n",
        "    1 text = 1 sequence vector with fixed length.\n",
        "    \"\"\"\n",
        "    # Create vocabulary with training texts.\n",
        "    tokenizer = text.Tokenizer(num_words=TOP_K)\n",
        "    tokenizer.fit_on_texts(train_texts)\n",
        "\n",
        "    # Vectorize training and validation texts.\n",
        "    x_train = tokenizer.texts_to_sequences(train_texts)\n",
        "    x_val = tokenizer.texts_to_sequences(val_texts)\n",
        "\n",
        "    # Get max sequence length.\n",
        "    max_length = len(max(x_train, key=len))\n",
        "    if max_length > MAX_SEQUENCE_LENGTH:\n",
        "        max_length = MAX_SEQUENCE_LENGTH\n",
        "\n",
        "    # Fix sequence length to max value. Sequences shorter than the length are\n",
        "    # padded in the beginning and sequences longer are truncated\n",
        "    # at the beginning.\n",
        "    x_train = sequence.pad_sequences(x_train, maxlen=max_length)\n",
        "    x_val = sequence.pad_sequences(x_val, maxlen=max_length)\n",
        "    return x_train, x_val, tokenizer.word_index,tokenizer,max_length"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "8VoeVu7MQFRj",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "# Build, Train, and Evaluate Your Model"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "dD6BOnEB7LBO",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "def _get_last_layer_units_and_activation(num_classes):\n",
        "    \"\"\"\n",
        "    Gets the number of output neuron  and \n",
        "    activation function for the last network layer.\n",
        "    \"\"\"\n",
        "    if num_classes == 2:\n",
        "        activation = 'sigmoid'\n",
        "        units = 1\n",
        "    else:\n",
        "        activation = 'softmax'\n",
        "        units = num_classes\n",
        "    return units, activation\n",
        "    "
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "uk3BpX_1RAcj",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "### Model Architecture\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "qYQ19nTf0RHt",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "def sequence_model(\n",
        "                 embedding_dim,\n",
        "                 dropout_rate,\n",
        "                 input_shape,\n",
        "                 num_classes,\n",
        "                 num_features\n",
        "                 ):\n",
        "    \"\"\"\n",
        "    Creates an instance of a sequence_model\n",
        "\n",
        "    # Arguments\n",
        "        embedding_dim: int, dimension of the embedding vectors.\n",
        "        dropout_rate: float, percentage of input to drop at Dropout layers.\n",
        "        input_shape: tuple, shape of input to the model.\n",
        "        num_classes: int, number of output classes.\n",
        "        num_features: int, number of words (embedding input dimension).\n",
        "    # Returns\n",
        "        A sequence model instance.\n",
        "    \"\"\"\n",
        "    op_units, op_activation = _get_last_layer_units_and_activation(num_classes) #get op_unit and output activation function\n",
        "\n",
        "    #Model Architecture\n",
        "    model = models.Sequential()\n",
        "    model.add(Embedding(input_dim=num_features,\n",
        "                            output_dim=embedding_dim,\n",
        "                            input_length=input_shape[0]))\n",
        "    model.add(GlobalMaxPool1D())\n",
        "    model.add(Dense(embedding_dim, activation = 'relu'))\n",
        "    model.add(Dropout(rate=dropout_rate))\n",
        "    model.add(Dense(op_units, op_activation))\n",
        "    return model"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "wJfLYKPBT9eM",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "### Number Of classes"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "Vmmp4XbR2a5w",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "def get_num_classes(labels):\n",
        "    \"\"\"\n",
        "    Gets the total number of classes\n",
        "    # Arguments\n",
        "        labels: label values\n",
        "    # Returns\n",
        "             total number of classes.\n",
        "    \"\"\"\n",
        "    num_classes = max(labels) + 1\n",
        "    missing_classes = [i for i in range(num_classes) if i not in labels]\n",
        "    if len(missing_classes):\n",
        "        raise ValueError('Missing samples with label value(s) '\n",
        "                         '{missing_classes}. Please make sure you have '\n",
        "                         'at least one sample for every label value '\n",
        "                         'in the range(0, {max_class})'.format(\n",
        "                            missing_classes=missing_classes,\n",
        "                            max_class=num_classes - 1))\n",
        "\n",
        "    if num_classes <= 1:\n",
        "        raise ValueError('Invalid number of labels: {num_classes}.'\n",
        "                         'Please make sure there are at least two classes '\n",
        "                         'of samples'.format(num_classes=num_classes))\n",
        "    return num_classes"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "bynQBGPHVk3X",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "## Data Generator\n",
        "genrate data in term of baches"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "QimWk_DLWCR6",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "def _data_generator(x, y, num_features, batch_size):\n",
        "    \"\"\"\n",
        "    Generates batches of vectorized texts for training/validation\n",
        "\n",
        "    # Arguments\n",
        "        x: feature matrix.\n",
        "        y:  labels.\n",
        "        num_features:  number of features.\n",
        "        batch_size: number of samples per batch.\n",
        "    # Returns\n",
        "        Yields feature and label data in batches.\n",
        "    \"\"\"\n",
        "    num_samples = x.shape[0]\n",
        "    num_batches = num_samples // batch_size\n",
        "    if num_samples % batch_size:\n",
        "        num_batches += 1\n",
        "\n",
        "    while 1:\n",
        "        for i in range(num_batches):\n",
        "            start_idx = i * batch_size\n",
        "            end_idx = (i + 1) * batch_size\n",
        "            if end_idx > num_samples:\n",
        "                end_idx = num_samples\n",
        "            x_batch = x[start_idx:end_idx]\n",
        "            y_batch = y[start_idx:end_idx]\n",
        "            yield x_batch, y_batch"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "zF_h0LsDWic-",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "## Train Model"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "rgaGOEXt0RnL",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "def batch_train_sequence_model(data,\n",
        "                               learning_rate=1e-3,\n",
        "                               epochs=70,\n",
        "                               batch_size=1000,\n",
        "                               dropout_rate=0.2,\n",
        "                               embedding_dim=32,\n",
        "                               ):\n",
        "    \"\"\"Trains sequence model on the given dataset.\n",
        "    # Arguments\n",
        "        data: tuples of training and test texts and labels.\n",
        "        learning_rate: learning rate for training model.\n",
        "        epochs: number of epochs.\n",
        "        batch_size:  number of samples per batch.\n",
        "        dropout_rate: percentage of input to drop at Dropout layers.\n",
        "        embedding_dim: dimension of the embedding vectors.\n",
        "    \"\"\"\n",
        "    # Get the data.\n",
        "    (train_texts, train_labels), (val_texts, val_labels) = data\n",
        "\n",
        "    # Verify that validation labels are in the same range as training labels.\n",
        "    num_classes = get_num_classes(train_labels)\n",
        "    unexpected_labels = [v for v in val_labels if v not in range(num_classes)]\n",
        "    if len(unexpected_labels):\n",
        "        raise ValueError('Unexpected label values found in the validation set:'\n",
        "                         ' {unexpected_labels}. Please make sure that the '\n",
        "                         'labels in the validation set are in the same range '\n",
        "                         'as training labels.'.format(\n",
        "                             unexpected_labels=unexpected_labels))\n",
        "\n",
        "    # Vectorize texts.\n",
        "    x_train, x_val, word_index,tokenizer,max_length = sequence_vectorize(\n",
        "            train_texts, val_texts)\n",
        "\n",
        "    # Number of features will be the embedding input dimension. Add 1 for the\n",
        "    # reserved index 0 for padding\n",
        "    num_features = min(len(word_index) + 1, TOP_K)\n",
        "\n",
        "    # Create model instance.\n",
        "    model = sequence_model( embedding_dim=embedding_dim,\n",
        "                                     dropout_rate=dropout_rate,\n",
        "                                     input_shape=x_train.shape[1:],\n",
        "                                     num_classes=num_classes,\n",
        "                                     num_features=num_features)\n",
        "\n",
        "    # Compile model with learning parameters.\n",
        "    if num_classes == 2:\n",
        "        loss = 'binary_crossentropy' # if number of classes is two then use this loss function else\n",
        "    else:\n",
        "        loss = 'sparse_categorical_crossentropy'\n",
        "    optimizer = tf.keras.optimizers.Adam(lr=learning_rate)\n",
        "    model.compile(optimizer=optimizer, loss=loss, metrics=['acc'])\n",
        "\n",
        "    # Create callback for early stopping on validation loss. If the loss does\n",
        "    # not decrease in two consecutive tries, stop training.\n",
        "    callbacks = [tf.keras.callbacks.EarlyStopping(\n",
        "        monitor='val_loss', patience=2)]\n",
        "\n",
        "    # Create training and validation generators.\n",
        "    training_generator = _data_generator(\n",
        "        x_train, train_labels, num_features, batch_size)\n",
        "    validation_generator = _data_generator(\n",
        "        x_val, val_labels, num_features, batch_size)\n",
        "\n",
        "    # Get number of training steps. This indicated the number of steps it takes\n",
        "    # to cover all samples in one epoch.\n",
        "    steps_per_epoch = x_train.shape[0] // batch_size\n",
        "    if x_train.shape[0] % batch_size:\n",
        "        steps_per_epoch += 1\n",
        "\n",
        "    # Get number of validation steps.\n",
        "    validation_steps = x_val.shape[0] // batch_size\n",
        "    if x_val.shape[0] % batch_size:\n",
        "        validation_steps += 1\n",
        "\n",
        "    # Train and validate model.\n",
        "    history = model.fit_generator(\n",
        "            generator=training_generator,\n",
        "            steps_per_epoch=steps_per_epoch,\n",
        "            validation_data=validation_generator,\n",
        "            validation_steps=validation_steps,\n",
        "            callbacks=callbacks,\n",
        "            epochs=epochs,\n",
        "            verbose=2)  # Logs once per epoch.\n",
        "\n",
        "    # Print results.\n",
        "    history = history.history\n",
        "    print('Validation accuracy: {acc}, loss: {loss}'.format(\n",
        "            acc=history['val_acc'][-1], loss=history['val_loss'][-1]))\n",
        "\n",
        "    # Save model.\n",
        "    model.save('model.h5')\n",
        "    return history['val_acc'][-1], history['val_loss'][-1],tokenizer,max_length"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "CAH9mdPaTPPc",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "data=((X_train, np.array(y_train)),(X_val, np.array(y_val)))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "7ALJ07Fy0RuT",
        "outputId": "a70b67fe-fe4b-47e1-9267-ed0792fb8b51",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow_core/python/keras/initializers.py:119: calling RandomUniform.__init__ (from tensorflow.python.ops.init_ops) with dtype is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Call initializer instance with the dtype argument instead of passing it to the constructor\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow_core/python/ops/resource_variable_ops.py:1630: calling BaseResourceVariable.__init__ (from tensorflow.python.ops.resource_variable_ops) with constraint is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "If using Keras pass *_constraint arguments to layers.\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow_core/python/ops/nn_impl.py:183: where (from tensorflow.python.ops.array_ops) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Use tf.where in 2.0, which has the same broadcast rule as np.where\n",
            "Epoch 1/70\n",
            "Epoch 1/70\n",
            "10/10 - 1s - loss: 0.6948 - acc: 0.4853 - val_loss: 0.6901 - val_acc: 0.5112\n",
            "Epoch 2/70\n",
            "Epoch 1/70\n",
            "10/10 - 0s - loss: 0.6864 - acc: 0.6670 - val_loss: 0.6846 - val_acc: 0.7257\n",
            "Epoch 3/70\n",
            "Epoch 1/70\n",
            "10/10 - 0s - loss: 0.6783 - acc: 0.7793 - val_loss: 0.6762 - val_acc: 0.7734\n",
            "Epoch 4/70\n",
            "Epoch 1/70\n",
            "10/10 - 0s - loss: 0.6647 - acc: 0.8409 - val_loss: 0.6624 - val_acc: 0.8043\n",
            "Epoch 5/70\n",
            "Epoch 1/70\n",
            "10/10 - 0s - loss: 0.6438 - acc: 0.8740 - val_loss: 0.6420 - val_acc: 0.8212\n",
            "Epoch 6/70\n",
            "Epoch 1/70\n",
            "10/10 - 0s - loss: 0.6133 - acc: 0.8878 - val_loss: 0.6142 - val_acc: 0.8315\n",
            "Epoch 7/70\n",
            "Epoch 1/70\n",
            "10/10 - 0s - loss: 0.5735 - acc: 0.8925 - val_loss: 0.5788 - val_acc: 0.8371\n",
            "Epoch 8/70\n",
            "Epoch 1/70\n",
            "10/10 - 0s - loss: 0.5246 - acc: 0.9010 - val_loss: 0.5374 - val_acc: 0.8464\n",
            "Epoch 9/70\n",
            "Epoch 1/70\n",
            "10/10 - 0s - loss: 0.4710 - acc: 0.9079 - val_loss: 0.4935 - val_acc: 0.8446\n",
            "Epoch 10/70\n",
            "Epoch 1/70\n",
            "10/10 - 0s - loss: 0.4158 - acc: 0.9138 - val_loss: 0.4506 - val_acc: 0.8521\n",
            "Epoch 11/70\n",
            "Epoch 1/70\n",
            "10/10 - 0s - loss: 0.3622 - acc: 0.9244 - val_loss: 0.4109 - val_acc: 0.8549\n",
            "Epoch 12/70\n",
            "Epoch 1/70\n",
            "10/10 - 0s - loss: 0.3121 - acc: 0.9339 - val_loss: 0.3758 - val_acc: 0.8614\n",
            "Epoch 13/70\n",
            "Epoch 1/70\n",
            "10/10 - 0s - loss: 0.2674 - acc: 0.9418 - val_loss: 0.3468 - val_acc: 0.8708\n",
            "Epoch 14/70\n",
            "Epoch 1/70\n",
            "10/10 - 0s - loss: 0.2286 - acc: 0.9497 - val_loss: 0.3239 - val_acc: 0.8736\n",
            "Epoch 15/70\n",
            "Epoch 1/70\n",
            "10/10 - 0s - loss: 0.1959 - acc: 0.9563 - val_loss: 0.3067 - val_acc: 0.8708\n",
            "Epoch 16/70\n",
            "Epoch 1/70\n",
            "10/10 - 0s - loss: 0.1678 - acc: 0.9633 - val_loss: 0.2951 - val_acc: 0.8699\n",
            "Epoch 17/70\n",
            "Epoch 1/70\n",
            "10/10 - 0s - loss: 0.1448 - acc: 0.9696 - val_loss: 0.2874 - val_acc: 0.8699\n",
            "Epoch 18/70\n",
            "Epoch 1/70\n",
            "10/10 - 0s - loss: 0.1258 - acc: 0.9728 - val_loss: 0.2823 - val_acc: 0.8680\n",
            "Epoch 19/70\n",
            "Epoch 1/70\n",
            "10/10 - 0s - loss: 0.1113 - acc: 0.9762 - val_loss: 0.2790 - val_acc: 0.8661\n",
            "Epoch 20/70\n",
            "Epoch 1/70\n",
            "10/10 - 0s - loss: 0.0981 - acc: 0.9802 - val_loss: 0.2776 - val_acc: 0.8680\n",
            "Epoch 21/70\n",
            "Epoch 1/70\n",
            "10/10 - 0s - loss: 0.0870 - acc: 0.9836 - val_loss: 0.2772 - val_acc: 0.8670\n",
            "Epoch 22/70\n",
            "Epoch 1/70\n",
            "10/10 - 0s - loss: 0.0772 - acc: 0.9844 - val_loss: 0.2774 - val_acc: 0.8689\n",
            "Epoch 23/70\n",
            "Epoch 1/70\n",
            "10/10 - 0s - loss: 0.0696 - acc: 0.9870 - val_loss: 0.2774 - val_acc: 0.8680\n",
            "Validation accuracy: 0.867977499961853, loss: 0.27738057076931\n"
          ]
        }
      ],
      "source": [
        "val_acc,val_loss,tokenizer,max_length=batch_train_sequence_model(data)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "mDi4K5b8ZEwr",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "## Testing\n",
        "converting test data into sequence vector"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 119
        },
        "id": "ZvQr1wAPlNMg",
        "outputId": "91f05cb7-1b21-4e26-cb4c-ea45c443b9c5",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "3620                الله الله توظيف التخصص شعريا شعور رائع\n",
              "7464                      محتاجة تفكير إيجابي و تشجيع كتير\n",
              "10457    ذكرت ارقام الوظائف المباشرة وغير المباشرة في م...\n",
              "6653     سعوديات نطلب اسقاط الولايه ٥ نوفمبر ٢٠١٨م من ا...\n",
              "1165     برت من الجيد و للاسف اني شاهدته قدام عيوني ينج...\n",
              "Name: tweets, dtype: object"
            ]
          },
          "execution_count": 55,
          "metadata": {
            "tags": []
          },
          "output_type": "execute_result"
        }
      ],
      "source": [
        "X_test.head()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "WCTRt7Wd0Q0K",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "x_test = tokenizer.texts_to_sequences(X_test)\n",
        "x_test = sequence.pad_sequences(x_test, maxlen=max_length)\n",
        "num_features = min(len(tokenizer.word_index) + 1, TOP_K)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Z8IBUASUmTtE",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "Convert X_test in baches"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "4VugTSIGdKdy",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "testing_generator = _data_generator( x_test, y_test,num_features,1000)\n",
        "steps_per_epoch = x_test.shape[0] // 1000\n",
        "if x_test.shape[0] % 1000:\n",
        "  steps_per_epoch += 1"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lvI4nsDumbMf",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "Load Model"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 445
        },
        "id": "2fTn48Ppb4Wm",
        "outputId": "ed534c91-5dc9-4c76-af3d-94ec52bf6c42",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow_core/python/ops/init_ops.py:97: calling GlorotUniform.__init__ (from tensorflow.python.ops.init_ops) with dtype is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Call initializer instance with the dtype argument instead of passing it to the constructor\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow_core/python/ops/init_ops.py:97: calling Zeros.__init__ (from tensorflow.python.ops.init_ops) with dtype is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Call initializer instance with the dtype argument instead of passing it to the constructor\n",
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "embedding (Embedding)        (None, 58, 32)            640000    \n",
            "_________________________________________________________________\n",
            "global_max_pooling1d (Global (None, 32)                0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 32)                1056      \n",
            "_________________________________________________________________\n",
            "dropout (Dropout)            (None, 32)                0         \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 1)                 33        \n",
            "=================================================================\n",
            "Total params: 641,089\n",
            "Trainable params: 641,089\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ],
      "source": [
        "# load model\n",
        "model = tf.keras.models.load_model('/content/model.h5')\n",
        "# summarize model.\n",
        "model.summary()\n",
        "# evaluate the model"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lrQkhTDmmeLE",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "Get the Score on Testing"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        },
        "id": "acuIUQSLbnmi",
        "outputId": "b3d02108-a834-4bc6-df1e-bad56b258df3",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "5/5 - 0s - loss: 0.2705 - acc: 0.8776\n",
            "acc: 87.76%\n"
          ]
        }
      ],
      "source": [
        "score = model.evaluate_generator(generator=testing_generator, steps=steps_per_epoch,verbose=2)\n",
        "print(\"%s: %.2f%%\" % (model.metrics_names[1], score[1]*100))"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ZkeVUm5GmmfW",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "Now check Our model on السعودية examples.\n",
        "\n",
        "First collect examples of السعودية"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "4IikvbMK8by3",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "Saudi_Arabia_index = X_test.str.contains('السعودية')\n",
        "Saudi_Arabia=X_test[Saudi_Arabia_index]\n",
        "labels_Saudia=y_test[Saudi_Arabia_index]\n",
        "Saudia=pd.concat([Saudi_Arabia,labels_Saudia],axis=1, ignore_index=True,sort=False )"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "DETBH6yMM2ZQ",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "Saudia.index=np.arange(0, len(Saudia))\n",
        "Saudia.columns=['tweets','labels']"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 204
        },
        "id": "PoWsngE9JmH3",
        "outputId": "c5774ff4-6df9-4140-b826-414be3cd27ab",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "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>tweets</th>\n",
              "      <th>labels</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>السعودية الجديدة وضعت لها خطط لتحقيق أهداف محد...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>سعوديات نطلب اسقاط الولايه الولاية على النساء ...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>معليك منها ياحامد في بعض العينات في السعودية ي...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>كل الشكر للأنظمة الجديدة و القوانين الصارمة لأ...</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>لا يرجد عدل في السعودية ولا توجد شريعة كذلك هذ...</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                                              tweets  labels\n",
              "0  السعودية الجديدة وضعت لها خطط لتحقيق أهداف محد...       1\n",
              "1  سعوديات نطلب اسقاط الولايه الولاية على النساء ...       1\n",
              "2  معليك منها ياحامد في بعض العينات في السعودية ي...       1\n",
              "3  كل الشكر للأنظمة الجديدة و القوانين الصارمة لأ...       1\n",
              "4  لا يرجد عدل في السعودية ولا توجد شريعة كذلك هذ...       0"
            ]
          },
          "execution_count": 62,
          "metadata": {
            "tags": []
          },
          "output_type": "execute_result"
        }
      ],
      "source": [
        "Saudia.head()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 68
        },
        "id": "h5k3exd1QxXz",
        "outputId": "a448aca8-b048-4749-d4ce-04a7bf38f004",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "tweets    object\n",
              "labels     int64\n",
              "dtype: object"
            ]
          },
          "execution_count": 63,
          "metadata": {
            "tags": []
          },
          "output_type": "execute_result"
        }
      ],
      "source": [
        "Saudia.dtypes"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "pMfAVsF1nEFU",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "Count Number of Exaples"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 68
        },
        "id": "h0FwX4RcNJIO",
        "outputId": "92a7380b-4e20-4428-c9ed-3ee7ba431e84",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "tweets    108\n",
              "labels    108\n",
              "dtype: int64"
            ]
          },
          "execution_count": 64,
          "metadata": {
            "tags": []
          },
          "output_type": "execute_result"
        }
      ],
      "source": [
        "Saudia.count()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "8QNfu3lvn-VV",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "Convert Data into Vectors"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "r_62to46b1qv",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "x_test = tokenizer.texts_to_sequences(Saudia.tweets)\n",
        "x_test = sequence.pad_sequences(x_test, maxlen=max_length)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "id": "9E9LpRPvNjQB",
        "outputId": "2aa7d3a8-c228-4b1e-825c-61fc49eab964",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "108"
            ]
          },
          "execution_count": 66,
          "metadata": {
            "tags": []
          },
          "output_type": "execute_result"
        }
      ],
      "source": [
        "len(x_test)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "jFJV6MXbolD-",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "Lets Check How Many We predict Correct and Wrong with Confusion Matrix"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "o-erHs6hZYFy",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "def getConfusionMatrix(y_hat,y_test):\n",
        "    TN = len([1 for ind in range(len(y_test)) if y_test.iloc[ind]==0 and y_hat.iloc[ind]==0])\n",
        "    FP = len([1 for ind in range(len(y_test)) if y_test.iloc[ind]==0 and y_hat.iloc[ind]==1])\n",
        "    FN = len([1 for ind in range(len(y_test)) if y_test.iloc[ind]==1 and y_hat.iloc[ind]==0])\n",
        "    TP = len([1 for ind in range(len(y_test)) if y_test.iloc[ind]==1 and y_hat.iloc[ind]==1])\n",
        "    return TN,FP,FN,TP"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "sm9i-LujoyvZ",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "y_hat = model.predict_classes(x_test)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 85
        },
        "id": "mzjnLtzdo6dM",
        "outputId": "0de9c2f7-7a5e-45c0-986f-411aaa103cf6",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "confusion matrix:\n",
            "\tpred-0\tpred-1\n",
            "is-0\tTN=34\tFP=11\n",
            "is-1\tFN=5\tTP=58\n"
          ]
        }
      ],
      "source": [
        "numTN,numFP,numFN,numTP = getConfusionMatrix(pd.Series(y_hat.reshape(-1)),Saudia.labels)\n",
        "print ('confusion matrix:')\n",
        "print ('\\tpred-0\\tpred-1')\n",
        "print ('is-0\\tTN=%d\\tFP=%d' %(numTN,numFP))\n",
        "print ('is-1\\tFN=%d\\tTP=%d' %(numFN,numTP))"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "at2aLeC5svgT",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "It means we predict All negative tweets correctly and 2 positive tweets as negative . Our concentration was only Negative which We dont want to predict Wrong. It means We get best Model."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "XFp77X6F-_Qw",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "## ROC Curve"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "MQc2tpGr-_Qx",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "from sklearn.metrics import roc_curve\n",
        "fpr_keras, tpr_keras, thresholds_keras = roc_curve(Saudia.labels, y_hat.ravel())"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "x38T876H-_Qy",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "from sklearn.metrics import auc\n",
        "auc_keras = auc(fpr_keras, tpr_keras)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "id": "OnjfVGut-_Q0",
        "outputId": "c0588c02-f9f4-463e-ada2-31da7666a1fa",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYIAAAEWCAYAAABrDZDcAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjIsIGh0\ndHA6Ly9tYXRwbG90bGliLm9yZy8li6FKAAAgAElEQVR4nOzdeXxM1/vA8U/2BQkJkiCi9iCofU/F\nkiYRS9GFtpTwpbSqVFFiqaoWRdUWipJoixYlSn35FV9LUrXEErtsSASRhWQms/z+UFMhkYhMJpl5\n3q9XXjIz9859zkyc595zzj3HTKvVahFCCGGyzA0dgBBCCMOSRCCEECZOEoEQQpg4SQRCCGHiJBEI\nIYSJk0QghBAmThKBEEKYOEtDByBEUfPx8eH27dtYWFhgb29Px44dmTp1KmXKlNFtc/z4cRYuXMjp\n06cxNzenZcuWjB8/ntq1a+u2ycjIYNGiRezZs4fU1FScnZ3p3LkzI0eOxMnJyRBFE0Iv5IpAGKXl\ny5dz4sQJtm7dyrlz5wgJCdG9duLECYYOHUqXLl04ePAge/fupV69erz11lvEx8cDoFQqGTRoEJcv\nX2bVqlX8/fff/Pzzz5QvX57Tp0/rLW6VSqW39xYiL5IIhFGrVKkSHTp0IDo6Wvfc3Llz6dWrF4MG\nDaJs2bKUL1+esWPH0qRJExYvXgzAtm3buHnzJt999x21a9fG3NwcZ2dnRo0ahbe3d67HunTpEu+9\n9x6tWrWiXbt2LF++HICJEyeyYMEC3XYRERF06tRJ99jHx4eQkBACAwNp2rQpISEhfPjhhznee9as\nWcyaNQuA9PR0Jk+eTIcOHejYsSMLFixArVYXzQcmTJIkAmHUEhMTOXjwINWrVwcgMzOTEydO8Oqr\nrz61rZ+fH4cPHwbg8OHDdOzYMUdz0rNkZGTw3nvv0bFjRw4ePMgff/xB27ZtCxxneHg4ISEhHDt2\njICAAPbv309GRgYAarWaXbt20aNHD+BhYrG0tOSPP/5g69atHDp0iE2bNhX4WEI8SRKBMEqjRo3i\n5ZdfxtvbGycnJ90ZdmpqKhqNhkqVKj21T6VKlUhJSQHg3r17uW6Tlz///JOKFSsyZMgQbGxsKFu2\nLE2aNCnw/u+88w5ubm7Y2tpStWpVGjRowH//+18Ajh49iq2tLU2bNuX27dvs37+fyZMnY29vj7Oz\nM4MHDyY8PLzAxxLiSdJZLIzSkiVLaNeuHZGRkYwbN46UlBQcHBxwcHDA3Nyc5ORkatWqlWOf5ORk\nKlSoAED58uVJTk4u8PFu3rypu+ooDDc3txyPe/TowY4dO+jduzc7duzQXQ3cuHEDlUpFhw4ddNtq\nNJqn9hfiecgVgTBqrVq14rXXXuOrr74CwN7enqZNm7Jr166ntv39999p06YNAO3ateN///sfDx48\nKNBx3NzcdB3NT7KzsyMrK0v3+Pbt209tY2ZmluOxn58fkZGRJCYmsmfPHgIDAwFwdXXF2tqao0eP\ncuzYMY4dO8bx48flikC8EEkEwugNGjSIw4cPc/78eQDGjRvH1q1bWbduHRkZGaSmprJgwQJOnjzJ\n6NGjAejVqxeurq588MEHXLlyBY1GQ0pKCsuXL2f//v1PHeOVV14hOTmZtWvXolQqycjI4NSpUwB4\nenqyf/9+7t27R3JyMj/88EO+MTs5OdGqVSsmTZpEtWrVdFcvlStXpn379syZM4eMjAw0Gg1xcXFE\nRkYW1cclTJAkAmH0nJyc6NWrF0uWLAGgRYsWrFq1ij179tCxY0c6d+5MdHQ0GzZsoEaNGgBYW1uz\ndu1aatasyZAhQ2jevDn9+/cnJSWFxo0bP3WMsmXLsnr1av7v//6P9u3b4+vrS0REBPAwqdSvXx8f\nHx+GDBmCv79/geLu0aMHhw8f1jULPfL111+TnZ2Nv78/LVu25MMPP3yuZiwhnmQmC9MIIYRpkysC\nIYQwcZIIhBDCxEkiEEIIEyeJQAghTFypu6Hs5MmT2NjYFGpfhUJR6H1LKymzaZAym4YXKbNCoaBp\n06a5vlbqEoGNjQ2enp6F2jc6OrrQ+5ZWUmbTIGU2DS9S5scnXnySNA0JIYSJk0QghBAmThKBEEKY\nOEkEQghh4iQRCCGEidNbIpg0aRJt27Z9asKsR7RaLbNmzaJbt24EBgZy9uxZfYUihBDiGfSWCF57\n7TVWrVqV5+sHDhwgJiaGP/74g88//5zp06frKxQhhBDPoLf7CFq2bElCQkKer+/du5fevXtjZmZG\n06ZNSUtL49atW1SuXFlfIQkhRImmVGlIzcwmNVNJamY29x5kk5qZTXLqfVJS0/GpYauX4xrshrKk\npCRcXV11j11dXUlKSso3ESgUimfeGPEsWVlZhd63tJIymwYpc8mh0Wq5r9SQrtCQodSQrlD/86+G\nDGXO35/cJkv1jFUB1Nl42LtQTg9lljuLjZyU2TRImYuWVqslM1utOyN/9O+TZ+qP/zx6Li0rm2et\n8mJrZU55O2sc7axwtLfF1dkKRzsrytv986+9FQ52VlhplKxduZQtP4XhUaUSK5cswtWlvF7uLDZY\nInBxcSExMVH3ODExERcXF0OFI4QwQv82tTxdiesq7sxs7ukqcyWpmSpSM5Vkq/OuzS3MzXQVt6O9\nFU5lrHmpYpnHnrP+t3K3/7eSd7CzwtbKIt+41Wo1Xl5eXLhwgfHjxzN9+nTs7Oz0dgVksETg4+ND\naGgoAQEBnDp1inLlykn/gBDiKRqNlvQs1cOK+rHK/MK1NPYlXn6sAv+3gn9UuT9Qqp/53uVsLXVn\n4Y52VtR3dcDhsTPzx8/UHR89tremjLUFZmZmRV7WO3fu4OTkhIWFBV988QXu7u60aNGiyI/zJL0l\ngo8//pjIyEhSUlLo1KkTH3zwASqVCoC33noLb29v9u/fT7du3bCzs2P27Nn6CkUIYWCPmlqePCNP\ne6Jyz6255dlNLbextTL/p8J+eBbu7mT/sOJ+7Ixc9/ixM/VytpZYWpSMW6m0Wi1hYWGMGTOGOXPm\nMGzYMPr06VNsx9dbIvjmm2+e+bqZmRnTpk3T1+GFEHqQrdY8s8388bPxJ8/QlWpNnu9rYW6mq6Ad\n7J5uanF4vBL/p2JPToiheeMGBWpqKcni4+MZMWIEO3fupE2bNrRv377YYyh1ncVCiBej0WhJV6hI\nfVR553ZG/iDn82n/PH8/v6YWG8vHmlCscHEoi6Ndzgr88Qr/0XNlbSyfu6lFfdey1CeBH3/8kf/8\n5z+o1WoWLlzI6NGjsbAo/jJJIhCiFNJqtWRla3SVdVRiJnGaxGdX7v88Ts/KRvOMUS02luaPtY9b\nU62CPeWr5t7U8nhzi0MJamopLSpUqEDr1q0JCQnhpZdeMlgckgiEMKBHTS1PtZk/eKx5RVe559fU\nclP326Omlkc/FewfNrXkPBu3fqwy/3fb0n6WXZKpVCoWLFiAUqnks88+49VXX8XX11cvHc/PQxKB\nEC/oyaaWJ8/I0x5vQ8/8Z3jiP6NcCtLU8ngTSl2Xsv9U2DmbW+4l36RxvVq65wrT1CL069SpUwwd\nOpS///6b119/Ha1Wi5mZWYn4niQRCMHTTS1Ptpc/Xonfe6DM0SGalvnsphZrS3PKP1ZpVy1vRwM3\nh3+bX+wf6xB9rLnleZpaoqPv4VnVsYg+DVGUFAoFs2bNYs6cOTg5ObFp0yb69u1bIhLAI5IIhFHJ\nVmu4l6nmSnJGjk7ORzcKParon2xuSX3w7FEt5mb8W0H/86+Hc5kcTSqOeTS3SFOLabt06RJfffUV\nAwYM4JtvvsHZ2dnQIT1FEoEocR41taQ9cWb+VCX+WGWe9k9l/29TS2yu713WxjJHRV2nclndLf3l\ncxnd8uhGorLWlpibl5wzOFGyZWRksG3bNgYOHEijRo04f/48NWvWNHRYeZJEIPTiUVOLrgJ/kLMp\n5d5TZ+TKHJV6QZpaHlXaVcvb0sDNQfc4M/UO9Wu652hqedT0YiWjWoSe7dmzh+HDhxMbG0uzZs3w\n9PQs0UkAJBGIfGSrNTlvEHqQs6nlUUWfW+WuVOXf1KKbl+WfppbczsifbG7Jr6klOlqFp2fVov4o\nhHimlJQUxo8fz+rVq6lbty779+8vNRMBSiIQOShUaoat+5srtzJIzcwmQ6F65vaPN7U42llRu3LZ\nnPOy5NbcIk0twsio1Wrat2/PxYsXmTRpEsHBwdja6mftAH2QRCBy2HUmkQMXk3m1oStVytvlrMSf\nmIRLmlqEqbt9+7ZukrjZs2dTvXp1mjVrZuiwnpskApFD6NFYPJztWTqwmZyxC5EHrVbL+vXr+eij\nj5gzZw7Dhw+nd+/ehg6r0OR0TuhcSEznr5gUBrauLklAiDzExsbi5+fHoEGD8PT0pFOnToYO6YVJ\nIhA6YRGxWFua06+5u6FDEaJECg0NpVGjRvzvf/9j8eLFHDx4kPr16xs6rBcmTUMCgPsKFb8ev06A\nlxtOZawNHY4QJVKlSpVo3749K1aswMPDw9DhFBlJBAKA307dIEOh4u021Q0dihAlRnZ2NvPnzyc7\nO5upU6fi6+tL9+7dS9T0EEVBmoYEWq2W0KOx1HctR7PqFQwdjhAlwokTJ2jdujWTJk3i3LlzaP9Z\nJs3YkgBIIhDAyfh7nL2RxsA2Hkb5Ry7E88jKymLy5Mm0bNmSGzdu8Msvv/Djjz8a9f8NSQSCsIg4\nylhb0OdluRtXiMuXLzNv3jzeffddoqOjee211wwdkt5JIjBx9x4o2X7qBr1frkpZG+kyEqYpIyOD\n9evXA9CoUSMuXLjA6tWrqVDBNJpKJRGYuM1/J6BQaRjY2nhGQAjxPHbv3k3Dhg0ZNGgQ0dHRAAZd\nNtIQJBGYMK1Wy4aIOJpVL0+DKg6GDkeIYnXnzh0GDRrEq6++ir29PQcPHiw1k8QVNWkLMGFHrtzh\n6u37fPN6E0OHIkSxejRJ3OXLl/nss8+YMmVKqZokrqhJIjBhoRGxlLe3wt/LzdChCFEskpOTcXZ2\nxsLCgq+++goPDw+aNm1q6LAMTpqGTNSttCz+OJtE/+bVZClFYfS0Wi1r1qyhbt26rFy5EoBevXpJ\nEviHJAIT9fNf8ag0WgZIJ7EwcjExMfj6+jJkyBC8vLzo3LmzoUMqcSQRmCC1RsuPkXF0qF2RlyqW\nMXQ4QujN+vXradSoEUeOHGHp0qX8+eef1K1b19BhlTiSCEzQ/52/xY3ULJlXSBg9FxcXOnXqxNmz\nZxk5ciTm5lLl5UY6i01QaEQsLg42dPF0MXQoQhSp7Oxsvv76a9RqNcHBwXTv3p3u3bsbOqwST9Kj\niYm/+4D9F5N5o2V1WWZSGJXjx4/TsmVLpkyZwoULF3STxIn8SU1gYjZExmEGvNVKFp8RxiEzM5OJ\nEyfSqlUrkpKS2LJlC2FhYUY9SVxR02siOHDgAL6+vnTr1o2QkJCnXr9x4wbvvPMOvXv3JjAwkP37\n9+szHJOnUKnZ+Fc8XTxdcHO0M3Q4QhSJq1ev8s033zB48GDOnTtXqtcONhS99RGo1WpmzpzJmjVr\ncHFxoV+/fvj4+FC7dm3dNsuWLcPPz48BAwZw+fJlhg8fzr59+/QVksnbdSaRO/eVvN1GhoyK0i0t\nLY0tW7bg6elJw4YNuXTpklGtGFbc9HZFEBUVhYeHB+7u7lhbWxMQEMDevXtzbGNmZkZGRgYA6enp\nVK5cWV/hCB5ON13dyZ6OtSsaOhQhCm3nzp00atSIqVOn6iaJkyTwYvR2RZCUlISrq6vusYuLC1FR\nUTm2GT16NEOHDiU0NJTMzEzWrFmT7/sqFArdl/+8srKyCr1vafWozLEpSiKv3WVIcycuXDhv6LD0\nypS/Z2OWkpLCnDlz2L59O7Vq1WL16tUARl/ux+nrezbo8NHw8HD69OnDkCFDOHHiBBMmTGDHjh3P\nHOtrY2NT6BkCo6OjTW52wUdl/mnbGawtzBnl1wznsjaGDkuvTPl7NlZqtZoGDRpw9epVgoODmTx5\nMlevXjXqMufmRb7nZyUQvSUCFxcXEhMTdY+TkpJwcck5bn3z5s2sWrUKgJdffhmFQkFKSgrOzs76\nCsskPVCq+PX4dfy9XI0+CQjjkpSURKVKlbCwsGDevHl4eHjQuHFjQ4dldPTWR+Dl5UVMTAzx8fEo\nlUrCw8Px8fHJsY2bmxtHjhwB4MqVKygUCpycnPQVksn67eQN0hUqBkonsSgltFot33//PfXq1dON\nOAwMDJQkoCd6uyKwtLQkODiYoKAg1Go1ffv2pU6dOixatIhGjRrRpUsXJk6cyJQpU1i7di1mZmbM\nmTNHxv4WMa1WS2hELPVcytHCwzSW3ROl29WrVxk2bBj79u3D29ubrl27Gjoko6fXPgJvb2+8vb1z\nPDdmzBjd77Vr1+ann37SZwgm7+IdBWeup/F5r4aSZEWJ98MPP/D+++9jYWHB8uXLGTZsmMwPVAxk\nriEjF34hDXtrC3q/XNXQoQiRrypVquDj48OyZcuoVq2aocMxGZIIjFjqg2wOXLvPa83dKWdrZehw\nhHiKUqlkzpw5aDQapk+fTrdu3ejWrZuhwzI5cs1lxH45noBCrWVga5luWpQ8f/31F82bN2fatGlc\nvXpVJokzIEkERkqr1RIWEUu9ijY0qupo6HCE0Hnw4AHjx4+nTZs2pKSk8Ntvv7Fu3TrpwzIgSQRG\n6ujVu1xJvk9APQdDhyJEDteuXWPx4sUMGzaMs2fPEhgYaOiQTJ70ERip0IhYHO2s6FRDlqIUhpea\nmsqvv/7Ke++9R8OGDbl8+TLu7jIVekkhVwRG6FZ6FrvPJNKveTVsLOUrFoYVHh5Ow4YNCQoK4vz5\nh/NcSRIoWaSWMEKbjiWg0kgnsTCs5ORkBg4cSI8ePahQoQJHjhyhfv36hg5L5EKahoyMWqNlQ0Qc\n7Ws7U7NSWaJvGzoiYYrUajUdOnTg2rVrzJgxg4kTJ2JtbW3osEQeJBEYmT8v3OL6vUw+CzCtWRlF\nyZCYmEjlypWxsLBg/vz51KhRg0aNGhk6LJEPaRoyMmERcVQuZ0O3Bi75byxEEdFoNKxYsYK6deuy\nYsUKAHr06CFJoJTINxFkZWWxYsUKgoODAYiNjZW1hUuo+LsP+L8Lt3izpTtWFpLjRfG4fPkyXbp0\nYcSIEbRs2RJfX19DhySeU761xeTJk9FqtRw/fhyAypUrs2DBAr0HJp7fj5FxmAFvtpJOYlE81qxZ\ng5eXF8ePH2flypX897//pWbNmoYOSzynfBNBTEwMI0aMwNLyYXeCnZ2d3ApeAilVGjYei8envgtV\nytsZOhxhIqpXr46vry/nzp0jKChI7g4upfLtLLa2tkahUOi+4Pj4eKysZAKzkmb32URuZyh5u41c\nDQj9USgUfPnll2g0GmbOnEmXLl3o0qWLocMSLyjfRDBy5EiCgoJITEzk008/5a+//mLWrFnFEZt4\nDqFHY3F3sqNTnUqGDkUYqYiICIYOHcrZs2cZNGgQWq1WrgCMRL6JwNvbW9cGCDBhwgRZU7iEuXwr\nnYhrd/n01fqYm8t/TFG07t+/z9SpU1m4cCFVq1Zlx44dBAQEGDosUYTy7SMYMmQITk5OdO3ala5d\nu+Ls7MyQIUOKIzZRQKFH47CyMOP1FrKQhyh6sbGxLF26lBEjRnD27FlJAkYozysCpVKJUqnk9u3b\nZGRk6J7PyMjgxo0bxRKcyN8DpYpfjifg18gN57I2hg5HGIl79+6xefNmgoKCaNCgAZcvX5YVw4xY\nnolgw4YNrFmzhjt37tCjRw/dSKGyZcvy5ptvFluA4tl2nLpJepaKt9t4GDoUYSS2bdvGyJEjuXXr\nFh06dKB+/fqSBIxcnolg8ODBDB48mLVr1zJ48OBiDEk8j9CIWOq6lKVljQqGDkWUcrdu3eLDDz/k\n559/pnHjxvz2228ySZyJyLezePDgwVy5coUrV66gUCh0z8tiEoYXlXCPqIRUZvRsKKM3xAtRq9W0\nb9+euLg4Zs2axYQJE2SYuAnJNxEsXbqUQ4cOcfXqVTp06MD//vc/mjdvLomgBAg7GoedlQV9mlU1\ndCiilLpx4waurq5YWFiwaNEiatSoQYMGDQwdlihm+Y4a+v3331m3bh2VKlVi7ty5bNu2jczMzOKI\nTTxDamY2205dp/fLVXCwlTM38Xw0Gg3Lli2jfv36LF++HAB/f39JAiYq30RgY2ODhYUFlpaWZGRk\nUKlSJa5fv14csYln+PV4AlnZGga2lk5i8XwuXrxI586def/992ndujV+fn6GDkkYWL5NQw0aNCAt\nLY2+ffvSt29fypYtS+PGjYsjNpEHrVZLWEQcTdzL06iqo6HDEaXI999/z+jRo7G1tWX16tUMHjxY\n+pfEsxOBVqtl9OjRODg4MHDgQDp06EBGRgYNGzYsrvhELiKu3eXyrQzm9pOELJ5PjRo18PPzY8mS\nJbi5uRk6HFFCPDMRmJmZMWTIEHbs2AGAh4c0Q5QEoUdjcbC1pEfjKoYORZRwCoWCzz//HIBZs2bJ\nJHEiV/n2EdSvX59z584VRyyiAJLTFew+m0i/5u7YWVsYOhxRgh0+fJimTZvyxRdfcPPmTZk+XuQp\n3z6C6Oho+vXrh7u7O/b29roZB7ds2VIc8YknbDwWT7Zay0CZblrkISMjg88++4zFixfj7u7Orl27\nZNUw8Uz5JoJly5YV+s0PHDjAF198gUajoX///gwfPvypbXbu3Ml3332HmZkZ9evXZ/78+YU+nrFT\na7RsiIijbU1nalUqa+hwRAkVFxfHihUrGDVqFLNnz6ZcuXKGDkmUcPkmgurVC3fmqVarmTlzJmvW\nrMHFxYV+/frh4+ND7dq1ddvExMQQEhLCjz/+iKOjI3fu3CnUsUzFgYvJXL+XyWR/T0OHIkqY1NRU\nQkJCGD58OA0aNODq1atUqSJ9SKJg9LbCeVRUFB4eHri7u2NtbU1AQAB79+7Nsc3GjRsZOHAgjo4P\nh0DKOgfPFno0lkrlbOje0MXQoYgSZMuWLQQGBvL+++9z4cIFAEkC4rnke0VQWElJSbi6uuoeu7i4\nEBUVlWObmJgYAN588000Gg2jR4+mU6dOz3xfhUJBdHR0oWLKysoq9L6GlpSRzb7zt3jDqzyXL14o\n8H6lucyFZSplTk5O5osvvuCPP/6gXr16LF++HI1GYxJlB9P5nh+nrzIXKBEkJiYSExNDmzZtUCqV\nqFQq7O3tX/jgarWa2NhY1q9fT2JiIm+//Tbbt2/HwcEhz31sbGzw9Cxc00h0dHSh9zW08N0XMDOD\nDwKaUfU5FqcvzWUuLFMos1qtpmfPnsTHxzN79mwCAgJM7kZPU/ien/QiZX5WAsm3aWjz5s2MHDmS\nKVOmAHD9+nXef//9fA/q4uJCYmKi7nFSUhIuLi5PbePj44OVlRXu7u7UqFFDd5Ug/qVUafjpr3g6\n16v8XElAGJ+EhAQ0Gg0WFhZ8++23nDx5kkmTJslMoeKF5JsI1q9fz88//0zZsg9Hqbz00ksF6tT1\n8vIiJiaG+Ph4lEol4eHh+Pj45Nima9euREZGAnD37l1iYmJwd3cvTDmM2h/nErmdoZDFZ0yYRqNh\n8eLF1K9fXzeSz8/PT9YLEEUi36YhGxsbrK2tdY/VanXB3tjSkuDgYIKCglCr1fTt25c6deqwaNEi\nGjVqRJcuXejYsSOHDh3C398fCwsLJkyYQIUKssDKk8KOxlGtgh2d6lYydCjCAM6fP09QUBCHDh3C\n19eXHj16GDokYWTyTQTNmjVj5cqVKBQKjh49yoYNG3jllVcK9Obe3t54e3vneG7MmDG6383MzJg0\naRKTJk16vqhNyOVbGRy5eodPfOthYS6Tg5maVatWMXr0aOzt7fnhhx945513ZJI4UeTybRr65JNP\nKFOmDDVr1mTdunW0adOGsWPHFkdsAgiLiMXKwow3WkqTmSmqVasWgYGBREdH8+6770oSEHqR7xXB\nn3/+Sf/+/RkwYEBxxCMek6lU88vfCbzayI2KZW0MHY4oBllZWcycOROA2bNn07lzZzp37mzgqISx\ny/eKYNeuXXTv3p1JkyZx4MCBAvcRiBe3PeoGaVkqBraWeYVMwaFDh2jatClffvklycnJMkmcKDb5\nJoK5c+eye/duOnfuzK+//krXrl0JDg4ujthMXtjRWOpULkvrl5wMHYrQo/T0dD744AM6duyIQqFg\n9+7drFy5UpqBRLEp0BQT1tbWdOnShT59+uDl5cWuXbv0HZfJO52QyqmEVAa2ri4VgpFLSEhg1apV\nfPDBB5w+fZru3bsbOiRhYvLtIzh06BA7d+7kyJEjNGvWjF69ejFv3rziiM2khUXEYmdlQZ9m1Qwd\nitCDO3fusHHjRkaOHImnpydXr16VFcOEweSbCH7++Wf8/f2ZOnUqtra2xRGTyUvLymbbyRv0bFIF\nRzu5Y9SYaLVafvnlF0aNGsXdu3fx8fGhXr16kgSEQeWbCL799tviiEM8Zsvx62Rmq+VOYiNz8+ZN\nRo0axZYtW2jevLlusjghDC3PRPD2228TGhpKy5Ytc7RRP1qh7NHUEKJoabVaQo/G0riaI17VHA0d\njigiarWajh07cv36db7++mvGjh2LpaXeJv8V4rnk+Ze4bt06AI4ePVpswQiIvHaXS7cy+Lqvac0k\naazi4+OpWrUqFhYWLFmyhJdeeom6desaOiwhcshz1JC5+cOXPvvsMywsLHL8fPbZZ8UWoKkJi4ij\nnK0lgU1kYZHSTK1W8+233+aYJM7X11eSgCiR8r02fbTi0SNqtZrTp0/rLSBTdjtDwe9nbjKwtQd2\n1haGDkcUUnR0NEOHDuXIkSP4+fkRGBho6JCEeKY8E0FISAgrV67k/v37tGrVCvi3f6Bv377FFqAp\n2Xgsnmy1lrfbyJ3EpVVISAgffPAB5cqVY/369QwcOFDuAxElXp6JYNiwYQwZMoT58+czfvx43fMW\nFnKmqg8ajZYNEXG0qelE7crlDB2OKKQ6derQp08fvv32WypXrmzocIQokDwTQWxsLDVq1KBXr15c\nunTpqddlQYyitf9SMgkpmXz6qnyupUlmZibTp0/HzMyMOXPmyCRxolR6ZtPQ7NmzdTMhPs7MzIyw\nsDC9BmZqwo7GUrGsDb4NXU3EeUQAACAASURBVA0diiigAwcOEBQUxKVLlxgxYoSu6VSI0ibPRDB7\n9mwANmzYUGzBmKrr9zLZd/4WI1+phbVlgaZ/EgaUlpbGxIkTWbZsGTVr1mTv3r1PLcMqRGmSb63z\nxx9/kJGRATy8ShgzZgznz5/Xe2Cm5KfIOLTAmy2lk7g0uHHjBmvXruXjjz8mKipKkoAo9fJNBIsX\nL6Zs2bIcP36c/fv3ExgYKNNQF6FstYaf/oqnc73KuDvZGzockYfbt2+zdOlS4GH/2LVr15g/fz5l\nypQxcGRCvLh8E8GjUUJ//vknb7zxBl27dkWpVOo9MFOx51wSyekKGTJaQmm1Wn7++WcaNGjARx99\nxMWLFwFwcXExcGRCFJ18E0GlSpWYMWMG4eHheHt7o1Qq0Wg0xRGbSQg9GkvV8nZ415WhhiXNjRs3\n6N27N2+++SYeHh78/fffcmewMEr53lm8cOFC9u/fz9tvv42joyNJSUmMGzeuOGIzeleSMzh85Q6f\n+NbDwlxGm5QkarWaTp06cf36debNm8eYMWNkkjhhtPL9yy5Tpgx16tQhMjKSyMhIWrRogbe3d3HE\nZvQ2RMRhaW7G6y3cDR2K+EdsbCzVqlXDwsKCpUuXUrNmTWrXrm3osITQq3ybhkJDQxkzZgw3b97k\n5s2bfPTRR3IPQRHIylaz+e8EfBu5UqmcjaHDMXlqtZpvvvkGT09P3SRx3bt3lyQgTEKBVijbtGmT\nbnTEf/7zH958800GDhyo9+CM2fZTN0jNzObt1rL4jKGdOXOGoUOHEhkZSY8ePejdu7ehQxKiWBXo\n7iUrK6tcfxeFFxYRR61KZWhT08nQoZi05cuX06xZM65evcqGDRv47bffqFZN1okWpiXfK4KePXvy\n+uuv0717d7RaLXv37pUzphd05noqJ+PvEdyjgUxJYCCPpoPw9PSkf//+LFy4kEqVKhk6LCEMIt9E\nMGzYMFq3bs3ff/+NmZkZ06dPp3FjWT3rRYRFxGJrZU7f5nLmWdwePHhAcHAwFhYWfPXVV3h7e8vg\nB2HyCtQ0ZG1tneNHFF5aVjbbTt6gZ5MqONpJM1tx+vPPP2ncuDHz588nIyMDrVZr6JCEKBHyTQTL\nli1j3Lhx3Lp1i6SkJMaPH8+KFSuKIzajtPXEdR4o1QyUTuJik5qayn/+8x/d9ND79u1jyZIl0iwn\nxD/ybRraunUrW7duxc7ODoARI0bQu3dv/vOf/+g9OGOj1WoJPRqLV1VHmriXN3Q4JuPmzZuEhoYy\nfvx4ZsyYgb29zOkkxOPyvSKoXLkyarVa91itVhd45aUDBw7g6+tLt27dCAkJyXO73bt3U69ePaNf\nC/lYbAoXkzJkXqFikJyczOLFi4GHk8TFxMQwd+5cSQJC5CLfKwJHR0cCAgLo0KEDZmZmHDp0CC8v\nL7788ksAJk2alOt+arWamTNnsmbNGlxcXOjXrx8+Pj5P3aCTkZHBunXraNKkSREUp2QLPRpLOVtL\nAptUMXQoRkur1bJjxw6++uor0tLS8PX1pW7dujIiSIhnyDcRPDmqoqAVdlRUFB4eHri7P5w+ISAg\ngL179z6VCBYtWsSwYcP4/vvvnyfuUudOhoLfTycyoHV17K1lzhp9iI+PZ+TIkYSHh9O6dWu+//57\nmSROiALIt0bq379/od44KSkJV9d/l110cXEhKioqxzZnz54lMTGRV155pcCJQKFQEB0dXaiYsrKy\nCr3vi9p05h5KtYY2lVTFGoMhy1ycVCoVAQEB3L59m3HjxjF48GDMzc1NouxgOt/z46TMRcdgp6Ya\njYY5c+bompgKysbGBk9Pz0IdMzo6utD7vgiNRst/t/9Jq5eceLVt8TaBGarMxSUmJgZ3d3csLCxY\nvXo1NWvWRKFQGHWZc2Ps33NupMzPv29e9LZArouLC4mJibrHSUlJORbzuH//PhcvXuTdd9/Fx8eH\nkydPMnLkSKPsMD5wKZm4uw94u40MGS0qKpWKefPm4enpqVs5rGvXrtSsWdPAkQlR+hT4ikCpVD7X\nzWReXl7ExMQQHx+Pi4sL4eHhzJ8/X/d6uXLliIiI0D1+5513mDBhAl5eXgU+RmkRFhGHcxlrXm3o\nmv/GIl9RUVEMHTqUY8eO0atXL/r27WvokIQo1fK9IoiKiiIwMJDu3bsDcP78eT7//PN839jS0pLg\n4GCCgoLw9/fHz8+POnXqsGjRIvbu3fvikZcSN+5lsjc6iddbumNtqbcLMJOxdOlSmjdvTmxsLD//\n/DNbtmyhShUZhSXEi8j3imDWrFksX76cUaNGAQ/HZD9+Jv8suc3jMmbMmFy3Xb9+fYHes7T5KTIO\nLTCgldw78CIeTRLXqFEj3nzzTRYsWEDFihUNHZYQRiHfRKDRaKhatWqO58zN5cy2ILLVGn76K55X\n6lbC3UluZCqM+/fvM2XKFCwtLZk7dy6dOnWiU6dOhg5LCKOSb43u5uZGVFQUZmZmqNVq1q5dS40a\nNYohtNLvv+eSuJWukHmFCmnv3r14eXmxcOFCFAqFTBInhJ7kmwimT5/OmjVruHHjBu3atePUqVNM\nnz69GEIr/UIjYqniaEvn+gWbkkM8dO/ePYKCgujatSuWlpYcOHCAb7/9ViaJE0JP8m0acnZ2ZsGC\nBcURi1G5mpzBoct3GNetLhbmUoE9j6SkJH766Sc+/fRTpk2bppvwUAihH/kmgilTpuR6JlaQkUOm\nbENEHJbmZrzRyt3QoZQKjyr/MWPGUK9ePWJiYqQzWIhikm8iaNeune53hULBnj17cHNz02tQpV1W\ntppNfyfg29CVyuVsDR1OiabVagkLC2PMmDFkZGTg7+9PnTp1JAkIUYzyTQT+/v45Hvfq1YsBAwbo\nLSBjEB51k9TMbAa2liGjzxIXF8eIESP4/fffadu2Ld9//z116tQxdFhCmJznnmsoISGB27dv6yMW\noxEaEUvNSmVoW8vZ0KGUWCqVildeeYVbt27x7bff8v7772NhYWHosIQwSfkmgpYtW+r6CDQaDY6O\njowbN07vgZVWZ2+kciLuHlN7NJBRLrm4evUqHh4eWFpasnLlSmrVqiXDkYUwsGcmAq1Wy7Zt23ST\nxZmbm0vllo+wiDhsLM3p26xq/hubEJVKxfz585k2bRpff/01H374IV26dDF0WEII8rmPwMzMjOHD\nh2NhYYGFhYUkgXykZ2Wz9cR1AptUobx9wSfoM3YnT56kdevWTJw4EX9//0KvcSGE0I98byirX78+\n586dK45YSr2tJ2/wQKmW6aYf891339GyZUuuX7/O5s2b+fXXX2XUmRAlTJ5NQyqVCktLS6Kjo+nX\nrx/u7u7Y29vrJv/asmVLccZZ4mm1WsKOxtKwigNNqjkaOhyDe/R30rhxYwYOHMg333yDk5OTocMS\nQuQiz0TQv39/tmzZwrJly4oznlLr79gUziem8+VrXibdhJaRkcFnn32GlZUV8+bNk0nihCgF8kwE\njyb4ql5dxsIXRFhEHOVsLOnZxHTnxv/jjz8YPnw4cXFxfPDBB7qrAiFEyZZnIrh79y5r1qzJc8f3\n3ntPLwGVRnfvKwmPusmbrdwpY2OwZaANJiUlhY8//pi1a9dSr149Dhw4QIcOHQwdlhCigPKstTQa\nDffv3y/OWEqtTcfiUao1JttJfOvWLTZv3sykSZMIDg7G1lam1RCiNMkzEVSqVInRo0cXZyylkkaj\nZUNkHK1qOFHXpZyhwyk2iYmJ/Pjjj4wdO1Y3SZyzs9xJLURplOfwUVkEpGD+d/k2sXceMLCNafSl\naLVafvjhBxo0aMCkSZO4dOkSgCQBIUqxPBPB2rVrizGM0iv0aCzOZax5tZGroUPRu5iYGF599VUG\nDx5MgwYNOHnypEwSJ4QRyLNpqHz58sUZR6l0MzWTvedvMaxjTWwsjXvCNJVKRefOnbl9+zZLlixh\nxIgRsna1EEbC9Ia4FKGfIuPRaLUMaGW8zUKXL1/mpZdewtLSktWrV1OzZk08PEyzU1wIYyWndIWU\nrdbw019xdKpTierO9oYOp8hlZ2cze/ZsGjZsyJIlSwDo3LmzJAEhjJBcERTS3uhbJKUpmNXb+CrG\n48ePM3ToUE6ePEn//v154403DB2SEEKP5IqgkMIiYnFztKVzvUqGDqVIffvtt7Rq1YrExER+/fVX\nNm7cqJuGXAhhnCQRFMK12/c5eOk2b7WqjqWFcXyEj4YLv/zyy7z77rucO3eOPn36GDgqIURxkKah\nQvgxMg4LczPebOlu6FBeWHp6OpMmTcLGxob58+fTsWNHOnbsaOiwhBDFyDhOZ4tRVraaTcfi6d7A\nhcoOpXsqhV27dtGoUSOWLl2KVquVmwiFMFGSCJ7TztM3SXmQXarnFbpz5w6DBg3Cz8+PMmXKcOjQ\nIb755huZKVQIEyWJ4DmFRcRRs2IZ2tUqvVMq3Llzhy1btjB16lROnDhB27ZtDR2SEMKA9JoIDhw4\ngK+vL926dSMkJOSp19esWYO/vz+BgYEMGjSI69ev6zOcFxZ9M42/Y1MY0Lp6qTt7vnnzJvPmzUOr\n1VK3bl1iY2OZOXMmNjY2hg5NCGFgeksEarWamTNnsmrVKsLDw9mxYweXL1/OsY2npye//PIL27dv\nx9fXl7lz5+ornCIRejQWG0tz+jWvZuhQCkyr1bJ69Wo8PT2ZOnWq7juoUKGCgSMTQpQUeksEUVFR\neHh44O7ujrW1NQEBAezduzfHNm3atMHOzg6Apk2bkpiYqK9wXliGQsXWE9fp0bgK5e2tDR1OgVy7\ndo2goCCGDh1KkyZNOHXqlEwSJ4R4it6GjyYlJeHq+u+MnC4uLkRFReW5/ebNmwu0tq1CoSA6OrpQ\nMWVlZRV63/ALadxXqmnvqin0exQnlUqFn58f9+7dIzg4mNdffx21Wl0qYn9RL/I9l1ZSZtOgrzKX\niPsItm3bxpkzZwgNDc13WxsbGzw9PQt1nOjo6ELtq9VqGbv7IA3cHOjTsWmJ7h+4dOkSNWvWxMLC\ngtDQUDQaDT4+PoYOq1gV9nsuzaTMpuFFyvysBKK3piEXF5ccTT1JSUm5TlVw+PBhli9fzrJly7C2\nLplNLsfj7nE+MZ2323iU2CSQnZ3NrFmzaNSoEd999x0Ar7zyCm5ubgaOTAhR0uktEXh5eRETE0N8\nfDxKpZLw8PCnzkzPnTtHcHAwy5YtK9ErXIUdjaWsjSW9mlYxdCi5OnbsGC1atGDq1Km89tprvPXW\nW4YOSQhRiuitacjS0pLg4GCCgoJQq9X07duXOnXqsGjRIho1akSXLl34+uuvefDgAWPGjAHAzc2N\n5cuX6yukQkm5r2TH6Zu80cKdMjYloiUth0WLFvHxxx/j6urKtm3b6Nmzp6FDEkKUMnqt2by9vfH2\n9s7x3KNKH0rHcpib/05AqdKUuDWJtVotZmZmtGjRgqFDh/L111/LqnJCiEIpeae4JYhGoyUsIpYW\nHhWo7+pg6HAASEtL49NPP8XW1pYFCxbQvn172rdvb+iwhBClmEwx8QyHrtwm5s6DEjOv0M6dO2nY\nsCEhISFYWlrKJHFCiCIhieAZwo7G4VTGGj8v1/w31qPbt2/z9ttvExAQgKOjI4cPH2bu3LkldgST\nEKJ0kUSQh8TULPZEJ9G/eTVsLC0MGktKSgrbt29n2rRpHD9+nNatWxs0HiGEcZE+gjz89Fccao2W\nAa0N00l8/fp1wsLC+OSTT6hTpw6xsbHSGSyE0Au5IsiFSq3hp8h4OtWthIdzmWI9tlarZeXKlTRo\n0IDp06dz5coVAEkCQgi9kUSQi73nb5GYlsXAYr4auHLlCl26dGH48OE0a9aMqKgoateuXawxCCFM\njzQN5SL0aCxujrZ0qV+52I6pUqno0qULd+/eZcWKFQQFBWFuLnlaCKF/kgieEHvnPgcv3WZs17pY\nWui/Ir5w4QK1atXC0tKSH374gVq1alGtWulZ70AIUfrJKecTNkTEYWFuxhst3fV6HKVSyYwZM/Dy\n8mLJkiXAwzuxJQkIIYqbXBE8JitbzcZj8XTzdMHV0VZvx4mMjGTo0KGcOXOGAQMGMHDgQL0dSwgh\n8iNXBI/ZdSaRlAfZep1XaOHChbRt21Z3b0BYWBgVK1bU2/GEECI/kggeE3o0lhrO9rSvVfQV86Pp\nIFq1asWwYcM4e/YsPXr0KPLjCCHE85KmoX+cT0zjWGwKn/l7Ym5edFM3pKamMmHCBOzs7Fi4cCHt\n2rWjXbt2Rfb+QgjxouSK4B9hR+OwtjSnX/Oi66zdvn07DRo0YNWqVdjY2MgkcUKIEkkSAXBfoWLL\niev08HKjQpkXXy4zOTmZAQMG0LNnT5ydnTl69ChfffWVTBInhCiRJBEAW09eJ0OhYmARTTedmprK\nzp07mTFjBseOHaNly5ZF8r5CCKEPJt9HoNVqCT0aR33XcjSrXvj5fOLj4wkNDWXixInUrl2b2NhY\nHB0dizBSIYTQD5O/IjgRf4/om2m83cajUE03Go2G5cuX07BhQ2bNmqWbJE6SgBCitDD5RBB6NJYy\n1hb0frnqc+976dIlfHx8GDlyJK1ateL06dMySZwQotQx6aahew+U7Ii6Sf/m1Shr83wfhUqlolu3\nbty7d4/vv/+e9957TzqDhRClkkkngs1/J6BUaZ5rTeLo6Gjq1KmDpaUl69evp1atWlSpUkWPUQoB\n2dnZJCQkkJWVlefr0dHRxRyVYUmZc2dra0u1atWwsrIq8PuabCLQaLSERcTR3KMCnm4O+W6vUCiY\nPXs2s2fPZu7cuXz00Ud07NixGCIVAhISEihXrhw1atTI9cozMzMTOzs7A0RmOFLmp2m1Wu7cuUNC\nQgIvvfRSgd/XZPsIjly9w7Xb9wu0+MzRo0dp1qwZM2fO5K233uKdd94phgiF+FdWVhbOzs7S/Cie\nyczMDGdn5zyvHPNisokg9GgsFeyt8Pdye+Z28+fPp127dqSnp7Nz507WrVuHs7NzMUUpxL8kCYiC\nKMzfiUkmgqS0LP44l0T/Fu7YWlnkuo1GowGgbdu2jBgxgjNnzuDn51ecYQohRLEwyUTw81/xqDVa\n3mr1dLPQvXv3GDp0KGPGjAGgXbt2LF26FAeH/PsRhDBmL7/8su73/fv34+vry/Xr14vt+B9++CHx\n8fHFdrznFR8fT//+/enWrRsfffQRSqXyqW2ys7P59NNPCQwMxM/PjxUrVgAP+yD79etHz549CQgI\n4Ntvv9Xtc+TIEfr06UOvXr0YPHgwsbGxAISGhrJ58+Yiid3kEoFKreHHyDg61qnISxXL5Hht69at\nNGjQgB9++IFy5crJJHFC5OLIkSPMmjWLlStXUrVqwe6/UalUL3TMS5cuoVarcXcv+MqBarX6hY75\nvObNm8fgwYPZs2cPDg4OuVbSu3btQqlUsn37dn799Vd+/vlnEhISsLa25ocffuC3335j69atHDx4\nkJMnTwIwffp05s2bx7Zt2/Dz82PZsmUA9O3bl9DQ0CKJ3eRGDe07f4ubqVlMC2yoe+7WrVuMHj2a\nTZs20bRpU3bs2EGzZs0MGKUQefvl7wQ2Hst5ZqzRaDA3L/x53est3OlbgJl3//rrL6ZMmUJISAjV\nqz+8or579y7Tpk3jxo0bAEyePJnmzZuzePFi4uLiiI+Pp0qVKnz88cdMmDCBzMxMAKZOnUqzZs24\ndesWY8eOJSMjA7VazfTp02nRokWO427fvp0uXbroHk+bNo2oqCiUSiW+vr58+OGHAPj4+ODn58fh\nw4cJCgrCy8uLGTNmkJKSgq2tLZ9//jm1atVi3759LFu2jOzsbMqXL8+8efNeaIEorVbL0aNHmT9/\nPgB9+vThu+++Y8CAATm2MzMzIzMzE5VKRVZWFlZWVpQtWxYzMzPKlHl4YqpSqVCpVDna+jMyMnT/\nVq5cGQA7OzuqVq1KVFQUjRs3LnTsYIKJICwiDhcHG7p6VtY9l5aWxp49e/jiiy/45JNPnmv8rRCm\nQqlUMmrUKNatW0etWrV0z3/xxRcMGjSIFi1acOPGDYYOHcrvv/8OwJUrV9iwYQO2trZkZmayZs0a\nbGxsiImJ4eOPP+bXX39lx44ddOjQgZEjR6JWq3WJ4nHHjx8nICBA93js2LHY2NhgbW3N4MGDOX/+\nPPXr1wegfPnybNmyBYBBgwYxY8YMatSowalTp5gxYwbr1q2jefPmbNy4ETMzMzZt2sSqVauYOHFi\njmNevXqVsWPH5vpZrF+/PkdzcUpKCg4ODlhaPqxSXV1dSUpKemo/X19f9u7dS4cOHcjKymLSpEmU\nL/9wjjO1Ws1rr71GXFwcAwYMoEmTJrrPd/jw4djY2FCmTBk2bdqke79GjRpx7NgxSQTP42Z6Ngcu\nJfOhTx1uXE9g/fr1TJ48mdq1axMXF0e5cuUMHaIQ+erbvNpTZ+/FMabe0tKSl19+mc2bNzNlyhTd\n84cPH+by5cu6xxkZGdy/fx94eIZua/tw/W+VSsXMmTM5f/485ubmxMTEAODl5cXkyZNRqVR07doV\nT0/Pp46dnJyMk5OT7vHvv//OTz/9hEajITk5mStXrugSgb+/PwD379/nxIkTuv4+QNdun5iYyNix\nY0lOTkapVFKt2tNXQzVr1mTbtm2F+qzyEhUVhbm5OQcPHiQtLY0BAwbQrl073N3dsbCwYNu2baSl\npTFq1CguXrxI3bp1Wbt2LSEhITRp0oRly5bx5Zdf8sUXXwDg7OzM1atXXzguvSaCAwcO8MUXX6DR\naOjfvz/Dhw/P8bpSqWTChAmcPXuW8uXLs2DBgly/kKKy82Ia5mZmKKP/j4avjUej0fDGG29Qu3Zt\nSQJC5MPc3JyFCxcyePBgli9fzogRI4CHzVIbN27ExsbmqX0eT05r166lYsWKbNu2DY1GozuLbdmy\nJaGhoezfv5+JEyfy3nvv0bt37xzvY2Njg0KhAB52yq5evZrQ0FBcXFyYOHGi7rXHj6nVanFwcMi1\nMp81axaDBw+mS5cuRERE8N133z21zfNcEVSoUIG0tDRUKhWWlpYkJibi4uLy1H47duygY8eOWFlZ\n4ezsTLNmzTh9+nSOvg8HBwdat27NwYMHqVixIufPn9ddHfj6+jJ69GjdtgqFItfP/XnprbNYrVYz\nc+ZMVq1aRXh4ODt27Mhx1gCwadMmHBwc2LNnD4MHD2bevHn6CgeFSs2uC6lYJ19g4pgRtG3blrNn\nz8okcUI8Bzs7O1asWMH27dt1TRQdOnRg/fr1um3ymgIhPT2dSpUqYW5uzrZt23SdudevX6dixYq8\n/vrr9O/fn7Nnzz61b61atYiLiwMenunb2dlRtmxZbt++zYEDB3I9XtmyZalWrZqumUqr1XL+/Hld\nLI8q6q1bt+a6/6Mrgtx+nhxFaGZmRuvWrdm9ezcAW7ZswcfH56n3dHNzIyIiAoAHDx5w6tQpatas\nyd27d0lLSwMe3jx4+PBhatasiYODA+np6Vy7dg14eHPr481yMTEx1K1bN9f4n4feEkFUVBQeHh64\nu7tjbW1NQEAAe/fuzbHNvn376NOnD/Aw0x05ckRvI3X2nUskIxtuH/2VNWvWsHv3bmrUqKGXYwlh\nzMqXL8+qVatYtmwZe/fu5bPPPuPMmTMEBgbi7+/Pjz/+mOt+AwYMYMuWLfTs2ZOrV69ib28PQGRk\nJL169aJ3797s3LmTd99996l9vb29dRVo/fr1adCgAb1792bcuHHPHNgxd+5cNm/erBuW+d///heA\n0aNHM2bMGF577TVdG/2L+uSTT1izZo1uMsr+/fsDsHfvXhYtWgTAwIEDuX//PgEBAfTr14/XXnuN\n+vXrc+vWLd59910CAwPp168f7dq1o3PnzlhaWjJr1iw+/PBDevbsyY4dO5gwYYLumMePHy+SNdDN\ntHqqeXft2sXBgwd1bVlbt24lKiqK4OBg3TY9evRg1apVuLq6AtC1a1c2btyYoy3wSSdPnizUpdD1\ntGxCD0QzuJUrLpUr57+DkcjKytK10ZoKYyxzdnY2derUyfN1rVZr1HceZ2VlMWzYMNauXYuFxcOb\nQI29zLl5vMznz59n/fr1ujr2cZcuXcp10Etu/S9QCjuLbWxs8izMs3gCVR2sCrVvaRYdHS1lNgLR\n0dHP7Aw29gnY7OzsGDNmDGlpabrZfo29zLl5vMwPHjzg448/zvUzsLJ6uq571qyleksELi4uJCYm\n6h4nJSU91Xni4uLCzZs3cXV1RaVSkZ6eToUKFfQVkhCiFJPZfnNq3759kb2X3voIvLy8iImJIT4+\nHqVSSXh4+FOdJz4+Prrxvrt376ZNmzYmd6knREHJne6iIArzd6K3RGBpaUlwcDBBQUH4+/vj5+dH\nnTp1WLRoka7TuF+/fty7d49u3bqxZs0axo8fr69whCjVbG1tuXPnjiQD8UyP1iN43j4yvfYReHt7\n4+3tneO5x2/usLGxyTG5khAid9WqVSMhIYHk5ORcX8/Ozja5O+KlzLl7tELZ8yh1ncVCmCIrK6tn\nrjhljB3k+ZEyFx2Tm31UCCFETpIIhBDCxEkiEEIIE6e3O4v1pbB3FgshhClTKBQ0bdo019dKXSIQ\nQghRtKRpSAghTJwkAiGEMHGSCIQQwsRJIhBCCBMniUAIIUycJAIhhDBxRpkIDhw4gK+vL926dSMk\nJOSp15VKJR999BHdunWjf//+JCQkGCDKopVfmdesWYO/vz+BgYEMGjSI69evGyDKopVfmR/ZvXs3\n9erV4/Tp08UYnX4UpMw7d+7E39+fgIAAxo0bV8wRFr38ynzjxg3eeecdevfuTWBgIPv37zdAlEVn\n0qRJtG3blh49euT6ularZdasWXTr1o3AwMBc13h+blojo1KptF26dNHGxcVpFQqFNjAwUHvp0qUc\n24SGhmqnTp2q1Wq12h07dmjHjBljiFCLTEHKfOTIEe2DBw+0Wq1WGxYWZhJl1mq12vT0dO2AAQO0\n/fv310ZFRRkg0qJTyCdGRwAACZRJREFUkDJfu3ZN26tXL+29e/e0Wq1We/v2bUOEWmQKUuYpU6Zo\nw8LCtFqtVnvp0iVt586dDRFqkYmMjNSeOXNGGxAQkOvrf/75p3bo0KFajUajPXHihLZfv34vfEyj\nuyKIiorCw8MDd3d3rK2tCQgI0K1/8Mi+ffvo06cPAL6+vhw5cqRUz/NekDK3adNGt6Rd06ZNc6we\nVxoVpMwAixYtYtiwYUZxN3pByrxx40YGDhyIo6MjAM7OzoYItcgUpMxmZmZkZGQAkJ6eTuVSviZ5\ny5Ytdd9fbvbu3Uvv3r0xMzOjadOmpKWlcevWrRc6ptElgqSkJFxdXXWPXVxcSEpKemobNzc34OEC\nOuXKlSMlJaVY4yxKBSnz4zZv3kynTp2KIzS9KUiZz549S2JiIq+88koxR6cfBSlzTEwM165d4803\n3+T111/nwIEDxR1mkSpImUePHs327dvp1KkTw4cPZ8qUKcUdZrF68jNxdXV95v/3gjC6RCCebdu2\nbZw5c4agoCBDh6JXGo2GOXPm8Omnnxo6lGKlVquJjY1l/fr1zJ8/n6lTp5KWlmbosPQqPDycPn36\ncODAAUJCQpgwYQIajcbQYZUqRpcIXFxccjR7JCUl4eLi8tQ2N2/eBEClUpGenk6FChWKNc6iVJAy\nAxw+fJjly5ezbNkyrK2tizPEIpdfme/fv8/Fixd599138fHx4eTJk4wcObJUdxgX9G/bx8cHKysr\n3N3dqVGjBjExMcUcadEpSJk3b96Mn58fAC+//DIKhaJUX+Hn58nPJDExMdf/78/D6BKBl5cXMTEx\nxMfHo1QqCQ8Px8fHJ8c2Pj4+bNmyBXg4oqRNmzaYmZkZItwiUZAynzt3juDgYJYtW1bq240h/zKX\nK1eOiIgI9u3bx759+2jatCnLli3Dy8vLgFG/mIJ8z127diUyMhKAu3fvEhMTg7u7uyHCLRIFKbOb\nmxtHjhwB4MqVKygUCpycnAwRbrHw8fFh69ataLVaTp48Sbly5V64X8Tolqq0tLQkODiYoKAg1Go1\nffv2pU6dOixatIhGjRrRpUsX+vXrxyeffEK3bt1wdHRkwYIFhg77hRSkzF9//TUPHjzQrRnt5ubG\n8uXLDRx54RWkzMamIGXu2LEjhw4dwt/fHwsLCyZMmFCqr3YLUuaJEycyZcoU1q5di5mZGXPmzCnV\nJ3Yff/wxkZGRpKSk0KlTJz744ANUKhUAb731Ft7e3uzfv59u3bphZ2fH7NmzX/iYMg21EEKYOKNr\nGhJCCPF8JBEIIYSJk0QghBAmThKBEEKYOEkEQghh4iQRiBLL09OTXr166X6eNUtsQkJCnrM1FrfT\np08za9YsACIiIjh+/LjutR9//JGtW7cWWyzR0dGlfjZOoX9Gdx+BMB62trZs27bN0GE8Ny8vL92N\na5GRkdjb29OsWTPg4TjwoqZSqbC0zP2/cnR0NGfOnMHb27vIjyuMhyQCUaokJCQwYcIEMjMzAZg6\ndaqukn3k0qVLTJo0iezsbDQaDYsXL6ZGjRps27aN9evXk52dTZMmTZg2bRoWFhY59vXx8eHVV1/l\n4MGD2NjYMH/+fDw8PEhISGDy5MmkpKTg5OTEl19+SZUqVfj9999ZsmQJ5ubmlCtXjrCwMCIiIli9\nejVTp07lp59+wtzcnN9++42pU6dy5MgR7O3teeWVV/j000/ZvHmzrlwjR45k+/btnDlzhjlz5vDg\nwQMqVKjAl19++dSdoxMnTsTa2pro/2/v/kKa3OM4jr/dRURdBXlXof0BrzIjkKCLrJR0e55NCoIu\nhCQ2EsHuVLoIFxX0lxZKQUREN1G6VtiIILwo6qKQICr/UKhBQhiJWJtbfboYjs7ZPBXnBEf3fd0N\nnt+f7bn47rdnfL6vXrFx40a8Xi9Hjx4lmUyyePFijh07xooVK4hEIiQSCZ49e0YoFGLr1q0cOXKE\noaEh0uk0zc3N7Nix4w/eMTMv/Osga2P+kLKyMrmuK9d11dTUJEn6/PmzEomEpEz2fn19vSRpbGws\nm98eDocVi8UkSclkUl++fNHw8LBCoZBmZmYkSYcPH1Y0Gs1Zs6qqSl1dXZKkaDSqYDAoSQqFQurp\n6ZEk3bhxQwcOHJAk+Xw+jY+PS5ImJyclSU+ePMmOi0QiunTpUnb+H1+7rqvR0VFJ0sWLF9XZ2amZ\nmRnt2bNHExMTkqTe3l61tbXl7LO1tVXBYFDpdFpSpu9CKpWSJD169EjNzc2SpO7ubnV0dGTHnT59\nWrdu3crut6amRtPT03PfBFMQ7ERg/rfy/TSUTqcJh8O8fv0aj8eTN1Btw4YNXLhwgfHxcWpqaigp\nKeHx48e8ePGC3bt3A5BIJObMXJp91uD1ejl+/DgA/f39nD9/HgC/38/JkyeBTMhZW1sbtbW1VFdX\n/9b7q62tJR6PEwwGicfjnD17lrdv3zI4OMi+ffuATIpqcXFx3vE7d+7MnmimpqZobW1lZGSEoqIi\nUqlU3jEPHz7kwYMHXL58GYBkMsn79+9Zs2bNb+3dLCxWCMy8cuXKFZYvX04sFuPbt2+sX78+5xrH\ncSgvL6evr49gMEhHRweSqK+v/89bN4bDYZ4/f05fXx+7du2iu7v7l8fW1dXR0tJCdXU1RUVFlJSU\nMDAwwLp167h+/fpPx882GoJMA57Kyko6Ozt59+4dDQ0Nc46LRCKsXr36l/dpFj7715CZV6ampigu\nLsbj8RCLxfj69WvONWNjY6xcuZKGhga2b9/OwMAAmzdv5t69e0xMTADw6dOnOfs2x+NxINP7t6Ki\nAsh88+/t7QXgzp07bNq0CYDR0VHKy8tpaWlh2bJlOZ3fli5dyvT0dN51Vq1ahcfjoaurKxujXFpa\nysePH+nv7wcglUoxNDT0S5/LbBTxbLJuvvW3bNnCtWvXsh35Xr58+dO5zcJnhcDMK3v37iUajeK6\nLm/evGHJkiU518TjcXw+H36/n8HBQQKBAGvXruXgwYM0NjbiOA6NjY18+PAh7xqTk5M4jsPVq1dp\nb28HMg+le3p6cByHWCzGoUOHADhx4gSO4+Dz+aioqKCsrOwvc1VVVXH//n38fj9Pnz7NWauuro7b\nt29nC8GiRYuIRCKcOnUK13UJBALZovBP9u/fz5kzZwgEAtmkSoDKykqGh4fx+/3cvXuXpqYm0uk0\nruvi9Xo5d+7cT+c2C5+ljxrzg23btnHz5s0FnWdvzN/ZicAYYwqcnQiMMabA2YnAGGMKnBUCY4wp\ncFYIjDGmwFkhMMaYAmeFwBhjCtx3wD3Hp69ULJsAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "plt.figure(1)\n",
        "plt.plot([0, 1], [0, 1], 'k--')\n",
        "plt.plot(fpr_keras, tpr_keras, label='Keras (area = {:.3f})'.format(auc_keras))\n",
        "plt.xlabel('False positive rate')\n",
        "plt.ylabel('True positive rate')\n",
        "plt.title('ROC curve')\n",
        "plt.legend(loc='best')\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Fw7qvQsttf4e",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "# Evaluate precision"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "LAW2O9DxpbJQ",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "def getPrecision(y_hat,y_test):\n",
        "    TN,FP,FN,TP=getConfusionMatrix(y_hat,y_test)\n",
        "    return TP/(TP+FP)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "id": "_B28FnYJt264",
        "outputId": "7957816c-11e2-47a5-ce3e-9707bcc98091",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "0.8405797101449275"
            ]
          },
          "execution_count": 74,
          "metadata": {
            "tags": []
          },
          "output_type": "execute_result"
        }
      ],
      "source": [
        "getPrecision(pd.Series(y_hat.reshape(-1)),Saudia.labels)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "W1elZ4RwtuTt",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": [
        "Evaluate Recall"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "xEyKzOtWti3n",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": [
        "def getRecall(y_hat,y_test):\n",
        "    TN,FP,FN,TP=getConfusionMatrix(y_hat,y_test)\n",
        "    return TP/(TP+FN)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "id": "XIDarU9uuYwk",
        "outputId": "acc07972-31e6-4da2-bc28-50c4b9a8bc24",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "0.9206349206349206"
            ]
          },
          "execution_count": 76,
          "metadata": {
            "tags": []
          },
          "output_type": "execute_result"
        }
      ],
      "source": [
        "getRecall(pd.Series(y_hat.reshape(-1)),Saudia.labels)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "uKf3gSUKwV8c",
        "pycharm": {
          "name": "#%% md\n"
        }
      },
      "source": []
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 122
        },
        "id": "wmz49MISwO52",
        "outputId": "4ba80511-318e-47a2-b41a-63474c701e62",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "سعوديات نطلب اسقاط الولايه الولاية على النساء وسلب ارادتهن استعباد للبشر بصك شرعي للولي يتصرف فيها كيف يشاء وين المنطق من الولاية الصهيونية اللي فرضت على المرأة السعودية  Predict: 1\n",
            "معليك منها ياحامد في بعض العينات في السعودية يعتبرون ان المرأة مالها حقوق وإن دافعت عن حقها او طالبت به فهي غير شريفة والمشكلة انهم مايتقبلوا اي سبب منطقي او عقلاني هم كذا نفس قوم قريش زمان يتبعون شي و  Predict: 0\n",
            "كل الشكر للأنظمة الجديدة و القوانين الصارمة لأمن الطرق و السلامة أخوي صار يربط حزام الأمان من تلقاء نفسه و الوالدة كذلك كلنا مع السعودية ايد واحدة  Predict: 1\n",
            "لا يرجد عدل في السعودية ولا توجد شريعة كذلك هذه الكذبة لم يعد يصدقها الا جاهل مئات المعتقلين بلا تهم او محاكمات حفلات ومجون هيئة الترفيه وانفتاح غير مبرر بنوك ربوية وغيره وغيره  Predict: 0\n",
            "عالميا يساوي الاستغلال الجنسي مليار دولار سنويا اليوم العالمي للمرأة السعودية مصر الامارات  Predict: 0\n"
          ]
        }
      ],
      "source": [
        "yhat=pd.Series(y_hat.reshape(-1))\n",
        "#checking first five\n",
        "for i in range(1,6): \n",
        "  print(Saudia.tweets[i],\" Predict:\",yhat[i])\n",
        "  "
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "0wcyAc7FzDx-",
        "pycharm": {
          "name": "#%%\n"
        }
      },
      "outputs": [],
      "source": []
    }
  ],
  "metadata": {
    "colab": {
      "provenance": []
    },
    "kernelspec": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.8.5"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}