Arabic_egy_sentiment_analysis.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Arabic_egy_sentiment_analysis.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "authorship_tag": "ABX9TyMrViYCHh3WeWQVOuNFr800",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/Amgad-Abdelkhaleq/Arabic-Egyptian-Sentiment-Analysis/blob/main/Arabic_egy_sentiment_analysis.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yXQ2x3ljf3iG",
        "outputId": "608fa7ba-66e1-4002-ff03-afdb5bf8cc96",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 102
        }
      },
      "source": [
        "import os\n",
        "import re\n",
        "import nltk\n",
        "import pickle\n",
        "import numpy as np\n",
        "import pandas as pd\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.metrics import accuracy_score\n",
        "from keras.preprocessing.text import Tokenizer\n",
        "from keras.preprocessing.sequence import pad_sequences\n",
        "from keras.models import Sequential\n",
        "from keras import layers\n",
        "from keras.backend import clear_session\n",
        "from keras.models import load_model\n",
        "nltk.download('punkt')\n",
        "nltk.download('wordnet')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[nltk_data] Downloading package punkt to /root/nltk_data...\n",
            "[nltk_data]   Package punkt is already up-to-date!\n",
            "[nltk_data] Downloading package wordnet to /root/nltk_data...\n",
            "[nltk_data]   Package wordnet is already up-to-date!\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "True"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 51
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YWxP8CWiPQXm",
        "outputId": "bd346c59-3a68-4c89-ac0f-0a9d1995e243",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "ar_stopwords = '''\n",
        "أنفسنا مثل حيث ذلك بشكل لدى ألا عن إلي ب لنا وقالت فقط الذي الذى ا هذا غير أكثر اي أنا أنت ايضا اذا كيف وكل أو اكثر أي أن منه وكان وفي تلك إن سوف حين نفسها هكذا قبل حول منذ هنا عندما على ضمن لكن فيه عليه قليل صباحا لهم بان يكون بأن أما هناك مع فوق بسبب ما لا هذه و فيها ف ولم ل آخر ثانية انه من الان جدا به بن بعض حاليا بها هم أ كانت هي لها نحن تم أنفسهم ينبغي إلى فان وقد تحت' عند وجود الى فأن الي او قد خارج إنه اى مرة هؤلاء أنها إذا فهي فهى كل يمكن جميع أنفسكم فعل كان ثم لي الآن وقال فى في ديك لم لن له تكون الذين ليس التى التي أنه وان بعد حتى ان دون وأن لماذا يجري كلا إنها لك ضد وإن فهو انها منها أى لديه ولا بين خلال وما اما عليها بعيدا كما نفسي نحو هو نفسك نفسه انت ولن إضافي لقاء وكانت هى فما أيضا إلا معظم ومن إما الا بينما وهي وهو وهى\n",
        "'''\n",
        "\n",
        "ar_stopwords=nltk.word_tokenize(ar_stopwords)\n",
        "print(\"length of stopwords is: \",len(ar_stopwords))\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "length of stopwords is:  179\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wf2a4Mge_UM9"
      },
      "source": [
        "pd.set_option('display.max_colwidth', None)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "78jg3Hgi_g2i"
      },
      "source": [
        "def process_text(text):\n",
        "    stemmer = nltk.ISRIStemmer()\n",
        "    word_list = nltk.word_tokenize(text)\n",
        "    #remove arabic stopwords\n",
        "    word_list = [ w for w in word_list if not w in ar_stopwords ]\n",
        "    #remove digits\n",
        "    word_list = [ w for w in word_list if not w.isdigit() ]\n",
        "    #stemming\n",
        "    word_list = [stemmer.stem(w) for w in  word_list]\n",
        "    return ' '.join(word_list) \n",
        "\n",
        "\n",
        "def clean_text(text):  \n",
        "\n",
        "    search = [\"أ\",\"إ\",\"آ\",\"ة\",\"_\",\"-\",\"/\",\".\",\"،\",\" و \",\" يا \",'\"',\"ـ\",\"'\",\"ى\",\n",
        "              \"\\\\\",'\\n', '\\t','&quot;','?','؟','!']\n",
        "    replace = [\"ا\",\"ا\",\"ا\",\"ه\",\" \",\" \",\"\",\"\",\"\",\" و\",\" يا\",\n",
        "               \"\",\"\",\"\",\"ي\",\"\",' ', ' ',' ',' ? ',' ؟ ', ' ! ']\n",
        "    #remove tashkeel\n",
        "    tashkeel = re.compile(r'[\\u0617-\\u061A\\u064B-\\u0652]')\n",
        "    text = re.sub(tashkeel,\"\", text)\n",
        "  \n",
        "    longation = re.compile(r'(.)\\1+')\n",
        "    subst = r\"\\1\\1\"\n",
        "    text = re.sub(longation, subst, text)\n",
        "    \n",
        "    text = re.sub(r\"[^\\w\\s]\", '', text)\n",
        "    #remove english words\n",
        "    text = re.sub(r\"[a-zA-Z]\", '', text)\n",
        "    #remove spaces\n",
        "    text = re.sub(r\"\\d+\", ' ', text)\n",
        "    text = re.sub(r\"\\n+\", ' ', text)\n",
        "    text = re.sub(r\"\\t+\", ' ', text)\n",
        "    text = re.sub(r\"\\r+\", ' ', text)\n",
        "    text = re.sub(r\"\\s+\", ' ', text)\n",
        "    #remove repetetions\n",
        "    text = text.replace('وو', 'و')\n",
        "    text = text.replace('يي', 'ي')\n",
        "    text = text.replace('اا', 'ا')\n",
        "    \n",
        "    for i in range(0, len(search)):\n",
        "        text = text.replace(search[i], replace[i])\n",
        "    \n",
        "    text = text.strip()\n",
        "    \n",
        "    return process_text(text) \n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "aX3eSJ4j_pXd",
        "outputId": "ccfbb4a5-ace0-465c-eaf4-7ee3a6fdcce4",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 197
        }
      },
      "source": [
        "data = pd.read_csv('DATA_SET_CP.csv')\n",
        "data.columns = ['text','sentiment']\n",
        "# exclude nutral sentiments\n",
        "data = data[data.sentiment.isin([-1, 1])]\n",
        "data.head()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>text</th>\n",
              "      <th>sentiment</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>عملي</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>اشتريت السلعه ودفعت ثمنها من الماستر كارد وتم ابلاغي انه تم ارجاعها ولا اعلم السبب</td>\n",
              "      <td>-1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>شنطة جميلة وأنيقة وسعرها مناسب جدا</td>\n",
              "      <td>-1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>حلوه</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5</th>\n",
              "      <td>هذا السرير (أو المحجر كما يسمى) ممتاز ... نوم ولعب بأمان خاصة للأطفال الرضع أقل من 3 سنوات.</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                                                                                          text  sentiment\n",
              "0                                                                                         عملي          1\n",
              "2           اشتريت السلعه ودفعت ثمنها من الماستر كارد وتم ابلاغي انه تم ارجاعها ولا اعلم السبب         -1\n",
              "3                                                           شنطة جميلة وأنيقة وسعرها مناسب جدا         -1\n",
              "4                                                                                         حلوه          1\n",
              "5  هذا السرير (أو المحجر كما يسمى) ممتاز ... نوم ولعب بأمان خاصة للأطفال الرضع أقل من 3 سنوات.          1"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 55
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "x5yW_169ETsf",
        "outputId": "e93fd8ec-e40f-4d99-d0ed-34bb6829967b",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 363
        }
      },
      "source": [
        "data['cleaned_text'] = data.text.apply(clean_text)\n",
        "data = data[data.cleaned_text != \"\"]\n",
        "data.head(10)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>text</th>\n",
              "      <th>sentiment</th>\n",
              "      <th>cleaned_text</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>عملي</td>\n",
              "      <td>1</td>\n",
              "      <td>عمل</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>اشتريت السلعه ودفعت ثمنها من الماستر كارد وتم ابلاغي انه تم ارجاعها ولا اعلم السبب</td>\n",
              "      <td>-1</td>\n",
              "      <td>شري سلع دفع ثمن استر كرد وتم بلغ رجع علم سبب</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>شنطة جميلة وأنيقة وسعرها مناسب جدا</td>\n",
              "      <td>-1</td>\n",
              "      <td>شنط جمل انق سعر نسب</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>حلوه</td>\n",
              "      <td>1</td>\n",
              "      <td>حله</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5</th>\n",
              "      <td>هذا السرير (أو المحجر كما يسمى) ممتاز ... نوم ولعب بأمان خاصة للأطفال الرضع أقل من 3 سنوات.</td>\n",
              "      <td>1</td>\n",
              "      <td>سرر حجر يسم متز نوم لعب بام خصه طفل رضع اقل سنو</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6</th>\n",
              "      <td>مفيده</td>\n",
              "      <td>1</td>\n",
              "      <td>فيد</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>7</th>\n",
              "      <td>مرااا حلوة احلى شيء انها تجمع اكتر من شيء بمكان واحد يعني اغتنام مساحهة ومنظمة ومرتبة وكل شي قريب من البيبي .. واجهت صعوبة بس في فتح السرير .. والطريقة كانت واضحة باليوتيوب اكتبو طريقة تركيب سرير الاطفال</td>\n",
              "      <td>1</td>\n",
              "      <td>مرا حله احل شيء جمع كتر شيء بمك وحد يعن غنم سحه نظم رتب شي قرب بيب وجه صعب بس فتح سرر طرق وضح تيب كبو طرق ركب سرر طفل</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8</th>\n",
              "      <td>هذه السلعه ممتازة جدا وفريدة من نوعها</td>\n",
              "      <td>1</td>\n",
              "      <td>سلع متز فرد نوع</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>10</th>\n",
              "      <td>تم توصيل السلعة الى المنزل بحالة ممتازه في وقت قصير وهي اداة مفيدة جدا لإراحة الأهل من تفقد الأطفال كل فترة</td>\n",
              "      <td>1</td>\n",
              "      <td>وصل سلع نزل بحل متز وقت قصر اده فيد ارح اهل فقد طفل فتر</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>11</th>\n",
              "      <td>منتج رائع جداً وانصح فيه</td>\n",
              "      <td>1</td>\n",
              "      <td>نتج رئع نصح</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                                                                                                                                                                                                           text  ...                                                                                                           cleaned_text\n",
              "0                                                                                                                                                                                                          عملي  ...                                                                                                                    عمل\n",
              "2                                                                                                                            اشتريت السلعه ودفعت ثمنها من الماستر كارد وتم ابلاغي انه تم ارجاعها ولا اعلم السبب  ...                                                                           شري سلع دفع ثمن استر كرد وتم بلغ رجع علم سبب\n",
              "3                                                                                                                                                                            شنطة جميلة وأنيقة وسعرها مناسب جدا  ...                                                                                                    شنط جمل انق سعر نسب\n",
              "4                                                                                                                                                                                                          حلوه  ...                                                                                                                    حله\n",
              "5                                                                                                                   هذا السرير (أو المحجر كما يسمى) ممتاز ... نوم ولعب بأمان خاصة للأطفال الرضع أقل من 3 سنوات.  ...                                                                        سرر حجر يسم متز نوم لعب بام خصه طفل رضع اقل سنو\n",
              "6                                                                                                                                                                                                         مفيده  ...                                                                                                                    فيد\n",
              "7   مرااا حلوة احلى شيء انها تجمع اكتر من شيء بمكان واحد يعني اغتنام مساحهة ومنظمة ومرتبة وكل شي قريب من البيبي .. واجهت صعوبة بس في فتح السرير .. والطريقة كانت واضحة باليوتيوب اكتبو طريقة تركيب سرير الاطفال  ...  مرا حله احل شيء جمع كتر شيء بمك وحد يعن غنم سحه نظم رتب شي قرب بيب وجه صعب بس فتح سرر طرق وضح تيب كبو طرق ركب سرر طفل\n",
              "8                                                                                                                                                                         هذه السلعه ممتازة جدا وفريدة من نوعها  ...                                                                                                        سلع متز فرد نوع\n",
              "10                                                                                                  تم توصيل السلعة الى المنزل بحالة ممتازه في وقت قصير وهي اداة مفيدة جدا لإراحة الأهل من تفقد الأطفال كل فترة  ...                                                                وصل سلع نزل بحل متز وقت قصر اده فيد ارح اهل فقد طفل فتر\n",
              "11                                                                                                                                                                                     منتج رائع جداً وانصح فيه  ...                                                                                                            نتج رئع نصح\n",
              "\n",
              "[10 rows x 3 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 56
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "w02VG47cEyoy",
        "outputId": "c188a2fb-bc96-43ef-ae63-d2626b55b029",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 409
        }
      },
      "source": [
        "min_sample = data.groupby(['sentiment']).count().text.min()\n",
        "input_data = pd.concat([data[data.sentiment == 1].head(min_sample), \n",
        "                        data[data.sentiment == -1].head(min_sample)])\n",
        "\n",
        "sns.set_style('darkgrid')\n",
        "plt.figure(figsize=(4, 5))\n",
        "sns.countplot(data['sentiment'], palette=['red', 'green'])\n",
        "plt.title('Number of messeges in with each sentiment')\n",
        "plt.xlabel(\"Negative(-1): Positive(1)\")\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/seaborn/_decorators.py:43: FutureWarning: Pass the following variable as a keyword arg: x. From version 0.12, the only valid positional argument will be `data`, and passing other arguments without an explicit keyword will result in an error or misinterpretation.\n",
            "  FutureWarning\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 288x360 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TZx0elLlFWPX"
      },
      "source": [
        "X = input_data.cleaned_text.values #training paramter\n",
        "Y = np.asarray(input_data.sentiment.values).astype('float32') #prediction paramter\n",
        "Y = Y.clip(0, 1)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LV0tHG4dGDw_",
        "outputId": "8f267831-9b6a-4ea3-b64c-a256b82f0657",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "tokenizer = Tokenizer(num_words=20000)\n",
        "tokenizer.fit_on_texts(X)\n",
        "X = tokenizer.texts_to_sequences(X)\n",
        "print(\"vocab size:\",len(tokenizer.word_index))\n",
        "X = pad_sequences(X, padding='post', maxlen=300)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "vocab size: 13286\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pLIElrWnHeKX",
        "outputId": "25b4e405-ebbc-41c4-b632-986862befe47",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        }
      },
      "source": [
        "X_train, X_test, label_train, label_test = train_test_split(X, Y, test_size=0.3, random_state=42)\n",
        "print(\"Training:\", len(X_train))\n",
        "print(\"Testing: \", len(X_test))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Training: 8328\n",
            "Testing:  3570\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "83lHsHqoHeME",
        "outputId": "c6dd81b0-e39f-4690-ec18-77fd57c99fe5",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 494
        }
      },
      "source": [
        "#create LSTM model with keras\n",
        "embedding_dim = 100\n",
        "dropout = 0.5\n",
        "opt = 'adam'\n",
        "clear_session()\n",
        "model = Sequential()\n",
        "model.add(layers.Embedding(input_dim=20000, \n",
        "                           output_dim=100, \n",
        "                           input_length=300))\n",
        "model.add(layers.Bidirectional(layers.LSTM(100, dropout=0.5, \n",
        "                                           recurrent_dropout=0.5, \n",
        "                                           return_sequences=True)))\n",
        "model.add(layers.GlobalMaxPool1D())\n",
        "model.add(layers.Dense(128, activation='relu'))\n",
        "model.add(layers.Dropout(dropout))\n",
        "model.add(layers.Dense(64, activation='relu'))\n",
        "model.add(layers.Dropout(dropout))\n",
        "model.add(layers.Dense(32, activation='relu'))\n",
        "model.add(layers.Dropout(dropout))\n",
        "model.add(layers.Dense(1, activation='sigmoid'))\n",
        "\n",
        "model.compile(optimizer=opt, \n",
        "              loss='binary_crossentropy', \n",
        "              metrics=['accuracy'])\n",
        "model.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "embedding (Embedding)        (None, 300, 100)          2000000   \n",
            "_________________________________________________________________\n",
            "bidirectional (Bidirectional (None, 300, 200)          160800    \n",
            "_________________________________________________________________\n",
            "global_max_pooling1d (Global (None, 200)               0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 128)               25728     \n",
            "_________________________________________________________________\n",
            "dropout (Dropout)            (None, 128)               0         \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 64)                8256      \n",
            "_________________________________________________________________\n",
            "dropout_1 (Dropout)          (None, 64)                0         \n",
            "_________________________________________________________________\n",
            "dense_2 (Dense)              (None, 32)                2080      \n",
            "_________________________________________________________________\n",
            "dropout_2 (Dropout)          (None, 32)                0         \n",
            "_________________________________________________________________\n",
            "dense_3 (Dense)              (None, 1)                 33        \n",
            "=================================================================\n",
            "Total params: 2,196,897\n",
            "Trainable params: 2,196,897\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pu9ljSuVI3Kv",
        "outputId": "da610c15-5859-429c-e5eb-4532891f099b",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 221
        }
      },
      "source": [
        "history = model.fit(X_train, label_train,\n",
        "                    epochs=4,\n",
        "                    verbose=True,\n",
        "                    validation_data=(X_test, label_test),\n",
        "                    batch_size=64)\n",
        "loss, accuracy = model.evaluate(X_train, label_train, verbose=True)\n",
        "print(\"Training Accuracy: {:.4f}\".format(accuracy))\n",
        "loss_val, accuracy_val = model.evaluate(X_test, label_test, verbose=True)\n",
        "print(\"Testing Accuracy:  {:.4f}\".format(accuracy_val))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/4\n",
            "131/131 [==============================] - 231s 2s/step - loss: 0.7016 - accuracy: 0.5328 - val_loss: 0.5637 - val_accuracy: 0.6798\n",
            "Epoch 2/4\n",
            "131/131 [==============================] - 233s 2s/step - loss: 0.3940 - accuracy: 0.8499 - val_loss: 0.2602 - val_accuracy: 0.8947\n",
            "Epoch 3/4\n",
            "131/131 [==============================] - 229s 2s/step - loss: 0.2198 - accuracy: 0.9293 - val_loss: 0.2300 - val_accuracy: 0.9104\n",
            "Epoch 4/4\n",
            "131/131 [==============================] - 229s 2s/step - loss: 0.1526 - accuracy: 0.9553 - val_loss: 0.2642 - val_accuracy: 0.9036\n",
            "261/261 [==============================] - 41s 156ms/step - loss: 0.0977 - accuracy: 0.9655\n",
            "Training Accuracy: 0.9655\n",
            "112/112 [==============================] - 17s 156ms/step - loss: 0.2642 - accuracy: 0.9036\n",
            "Testing Accuracy:  0.9036\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "eKNmUoriyrT8"
      },
      "source": [
        "def plot_history(history):\n",
        "    accuracy = history.history['accuracy']\n",
        "    val_accuracy = history.history['val_accuracy']\n",
        "    loss = history.history['loss']\n",
        "    val_loss = history.history['val_loss']\n",
        "    \n",
        "    epochs = range(1,len(accuracy) + 1)\n",
        "    \n",
        "    # Plot accuracy  \n",
        "    plt.figure(1)\n",
        "    plt.plot(epochs, accuracy, 'b', label='Training accuracy')\n",
        "    plt.plot(epochs, val_accuracy, 'g', label='Validation accuracy')\n",
        "    plt.title('Training and validation accuracy')\n",
        "    plt.xlabel('Epochs')\n",
        "    plt.ylabel('Accuracy')\n",
        "    plt.legend()\n",
        "    \n",
        "    # Plot loss\n",
        "    plt.figure(2)\n",
        "    plt.plot(epochs, loss, 'b', label='Training loss')\n",
        "    plt.plot(epochs, val_loss, 'g', label='Validation loss')\n",
        "    plt.title('Training and validation loss')\n",
        "    plt.xlabel('Epochs')\n",
        "    plt.ylabel('Loss')\n",
        "    plt.legend()\n",
        "    plt.show()\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Dyfzyef08yRh",
        "outputId": "a037b81c-4088-42dc-b235-f63b4047efd3",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 584
        }
      },
      "source": [
        "plot_history(history)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYkAAAEcCAYAAAAydkhNAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOzdd3hU1dbA4d+09N4bLUBCCISEjoBSBSGAChgFFEWqIirqFa9KE683+tkogkoXC8WCRBAVUXrvhJ6ElpBAGullZr4/AnOJEDIJycwkWe/z8JCZOWfP2jPJrDl7n7O2Qq/X6xFCCCHuQGnuAIQQQlguSRJCCCHKJElCCCFEmSRJCCGEKJMkCSGEEGWSJCGEEKJMkiSEyYwePZoff/yxyrc1px49erBjx44qbzc4OJjz588DMHXqVObNm2fUthX1888/M2rUqErteze7d+/m/vvvr/J2hempzR2AsGwRERGGn/Py8rCyskKlUgEwY8YMBg4caHRbCxcurJZta7uZM2dWSTuXLl2iZ8+eHD9+HLW65E9/4MCBFXoPRd0jSULc1cGDBw0/9+jRg1mzZnHffffdtl1xcbHhg0cIUXvIcJOolJvDCV988QWdO3fmjTfeIDMzk3HjxtGxY0fatWvHuHHjuHLlimGfJ598ktWrVwPwww8/8MQTTxAdHU27du3o0aMHf//9d6W2vXjxIsOHDyciIoKnn36aGTNm8Oqrr94xbmNi/OSTT3j88ceJiIhg1KhRpKWlGR7/6aef6N69Ox06dGD+/Pllvj6HDx+mc+fOaLVaw32///47AwYMAODIkSNERUXRtm1bunTpwsyZMyksLLxjW1OmTOHjjz823F64cCFdunShS5curFmzptS2f/31Fw8//DCtW7fmgQceYM6cOYbHRowYAUC7du2IiIjg4MGDhtf2pgMHDjB48GDatGnD4MGDOXDggNGvzd2cO3eOJ598krZt29K/f382bdpkeOzvv/+mX79+RERE0LVrVxYtWgRAWloa48aNo23btrRv355hw4ah0+mMej5RdSRJiEq7du0amZmZbN68mXfeeQedTsejjz7K5s2b2bx5M9bW1ncdKjly5AiNGjVi165djB49mjfffJOyqsTcbdtXX32VsLAwdu/ezcSJE1m7dm2Zz2lMjDExMbz33nvs3LmToqIiFi9eDMDZs2eZMWMG77//Plu3biUjI6NUgrlVq1atsLW1ZdeuXYb71q1bZ0gSSqWSN954g127dvHdd9+xc+dOvvnmmzLjvmnLli0sXryYxYsX89tvv7Fz585Sj9va2hIdHc2+ffv4/PPP+fbbb/njjz8AWLFiBQB79+7l4MGDpYYSATIyMhg3bhxPPvkku3fv5plnnmHcuHGkp6eX+9rcTVFREePHj6dz587s2LGDt956i1dffZW4uDgA3nzzTWbOnMnBgweJiYmhY8eOACxZsgRvb2927tzJ9u3bmTx5MgqFotznE1VLkoSoNKVSyaRJk7CyssLGxgZXV1f69OmDra0tDg4OTJgwgb1795a5v5+fH4899hgqlYpHHnmEq1evcu3atQptm5iYyNGjRw1xtG3blh49epT5nMbE+Oijj9KoUSNsbGzo27cvJ06cAODXX3+lW7dutGvXDisrK1588UWUyrL/hPr3709MTAwA2dnZbNmyhf79+wPQokULwsPDUavVBAQEEBUVddfX6qYNGzbw6KOPEhQUhJ2dHRMnTiz1eIcOHQgODkapVNKsWTP69+/Pnj17ym0XSo5CGjRowMMPP4xarSYyMpLAwEA2b95c7mtzN4cPHyY3N5exY8diZWVFp06d6N69O7/88gsAarWas2fPkp2djbOzM6GhoYb7r169SmJiIhqNhrZt20qSMAMZRBaV5urqirW1teF2Xl4e7733Hlu3biUzMxOAnJwctFqtYbL7Vh4eHoafbW1tAcjNzb3jc5W1bXp6Os7Ozob7AHx9fUlKSrpjO8bE6OnpWeq5bsaUkpKCj4+P4TE7OztcXFzu+DwAAwYM4PHHH2fGjBn8/vvvNG/eHH9/fwDi4+P573//y7Fjx8jLy0Or1Ro+HO8mJSWFFi1aGG7fbO+mw4cP83//93+cOXOGoqIiCgsL6du3b7nt3mzbz8+v1H1+fn4kJycbbpf12pTXro+PT6mEemu7s2fPZv78+Xz44YcEBwfzyiuvEBERwbPPPsvcuXMNZ19FRUUxduxYo/oiqo4cSYhK++e3usWLFxMfH8+qVas4cOAAX3/9NUCZQ0hVwdPTk8zMTPLy8gz3lZUg7jVGLy+vUsNLeXl5ZGRklLl9kyZN8PPzY8uWLcTExBAZGWl4bPr06QQGBrJx40YOHDjAyy+/bHQMt/YvMTGx1OOvvPIKPXv25O+//2b//v08/vjjhnbL+xbu5eV1W3tJSUl4e3uXG1d57V65cqXUfMKt7YaFhTF//nx27NhBr169eOmllwBwcHBgypQpbNq0ifnz57NkyZLbhtdE9ZMkIapMTk4O1tbWODk5kZGRwdy5c6v9Of39/WnRogVz5syhsLCQgwcPlhoeqcoY+/Tpw19//cW+ffsoLCxk9uzZ5U6kRkZGsmzZMvbu3VvqG31OTg729vbY29tz7tw5vv32W6Ni6Nu3Lz/++CNnz54lLy/vtvhzcnJwdnbG2tqaI0eOGIa7ANzc3FAqlVy8ePGObT/wwAMkJCSwbt06iouLWb9+PWfPnqVbt25GxVaWsLAwbGxsWLhwIUVFRezevZs///yTfv36UVhYyM8//0xWVhYajQZ7e3vDEcfmzZs5f/48er0eR0dHVCqVDDeZgSQJUWVGjhxJQUEBHTt2JCoqiq5du5rkef/v//6PQ4cO0aFDBz755BP69euHlZVVlcfYtGlTpk6dyquvvkrXrl1xcnIqNfx0J5GRkezdu5eOHTvi5uZmuP/1118nJiaG1q1b8/bbb9OvXz+jYnjggQcYOXIkI0eOpHfv3oZJ3pumTZvG7NmziYiIYN68eTz00EOGx2xtbRk/fjxPPPEEbdu25dChQ6X2dXV1ZcGCBSxZsoQOHTqwcOFCFixYUCruyrCysmLBggVs2bKFjh07Gib/GzduDMDatWvp0aMHrVu35rvvvuODDz4A4Pz58zzzzDNEREQQFRXFE088cVt/RfVTyKJDorZ56aWXCAwMZNKkSeYORYgaT44kRI135MgRLly4gE6nY8uWLWzatIlevXqZOywhagU5u0nUeNeuXeOFF14gIyMDHx8fpk+fTvPmzc0dlhC1ggw3CSGEKJMMNwkhhCiTJAkhhBBlkiQhhBCiTLVu4jo9PQedruLTLO7uDqSmZldDRKYnfbE8taUfIH2xVJXti1KpwNXVvszHa12S0On0lUoSN/etLaQvlqe29AOkL5aqOvoiw01CCCHKJElCCCFEmWrdcJMQ4t7o9XrS069SWJgP3D58kZKirDUrxNWdviiwsrLB1dWzwkUSJUkIIUrJzs5EoVDg7R2AQnH7YINaraS4uHZ8sNaVvuj1OjIyrpGdnYmjY9lroNyx3aoIzhjx8fFMmTKFjIwMXFxciI6OpmHDhqW2+de//sWpU6cMt0+dOsW8efPo2bOnqcIUos7Ly8vGzc37jglC1EwKhRJHR1fS0pItN0lMmzaNYcOGMWjQINauXcvUqVNZvnx5qW3ef/99w88nT55k5MiRJis3LYQoodNpUalkkKG2UanU6HTaCu9nkq8KqampxMbGGlbmioyMJDY2lrS0tDL3WbNmDQMGDChzXYCqFJuQxtj3/uDytZxqfy4hagJZ3Kf2qex7apKvCzeXKry5hrBKpTIsw3inBU0KCwtZt24dS5curfBzubs7VHif5mo1+b+c4NPVh4l+oSternYVbsPSeHo6mjuEKlNb+lJT+pGSokStvvv3x/IeryqjRj1FUVEhRUVFXLx4gcDAkoWKgoKCefvtGUa18cMPaygoyOeJJ0bc8fGbfdmy5W8OHz7ACy+8XDXBAxMmjGH48Cfp0uX+Kmvzbsp7X5RKZYV/Dy3ymPKPP/7Az8+PkJCQCu+bmppdqQtKZoztxOtzt/HmZ9t5Y0RrHO2q/wimunh6OnL1apa5w6gStaUvNakfOp3urpO5ppzs/eKLpQAkJSUyevSTLFnyjeGxmzEUFxejVpf9UTZw4KOltr/VrX25776u3Hdf1yrtm16vR6vVm+T1MuZ90el0t/0eKpWKu365NkmS8PX1JTk5Ga1Wi0qlQqvVkpKSgq+v7x23//777xk8eLApQjNo5OfMi0PC+HDlIT5ZfZhXH4/A1toic6gQdd6QIQPo2fNBDhzYS2BgE8aOfY7p098kJyeHwsJC7ruvM8899yIAixZ9Tl5eHhMnvsT69ev4/fdfcXR0Ii7uHI6OjsyaFY27uwfr169jx46tzJr1PgcO7GP27I9o3jyU48ePAgpmzPgPDRs2AuDzz+fx55+/4+TkTEREG/bv38uiRV/dNea0tFQ++OA9EhMvodfreeKJJ3nooUh0Oh0fffQ+Bw7sRaOxws7OlvnzF5Oensb06W+Rnp4KQNu27Zk06ZVqfV3vxCSfgu7u7oSEhBATE8OgQYOIiYkhJCTkjkNNV65cYf/+/Xz00UemCK2UoHoujB8UyrwfjjHvx6O8OKQVGhMdVgthqbYfTWLbkSTDbYUCqmoVmi5hvnRueecvi+XJycnhyy9LTn4pKCggOvpj7OzsKC4uZvLkiezatYOOHe+7bb8TJ2JZtuxbvL19eP/9d1mzZiXjxj1/23bx8ef497+n8q9/vcmyZYtYtmwR06bNYtu2LezYsY2lS7/F2tqat9563ah4P/nk/wgMbMx77/0f165d49lnRxAc3Izi4mIOHtzHihWrUSqVXL9+HYDfftuAv78/n376GYDhflMz2Sfg9OnTWbFiBX369GHFihXMmFEynjhmzBiOHj1q2O7HH3+ke/fuODs7myq0UiKaevL0Q82ITUhnYUxsrarrIkRt0rdvf8PPOp2Ozz77lJEjn+DZZ0cQF3eOM2dO33G/sLBWeHv7ANCiRUsSEy/dcbv69RsQFNQMgNDQlly+XLLdwYP76NGjF7a2tiiVSh56qP8d9/+nffv2MGhQydCXh4cHnTp15sCBffj5BVBcXMx///sOv/76i2H70NCW7Nq1g3nzPmX79q3Y2ZlnrtRk4ymNGzdm9erVt93/5Zdflro9YcIEU4VUpi5hvmTnFbFq81kc7DSM6B0kZ3uIOqtzy9Lf9i3lAjQ7O1vDzytXfk1W1nW++GIp1tbWREe/S2FhwR33u/WMSaVSiVZ759NCraysjdruXjk4OPDVV6s4eHA/+/btYf78OSxevIIWLcJYsuRr9u7dzcaN61mxYinz5y+qlhjuRsZSytC3Q336dqjP5gOXWbst3tzhCCHuIisrC3d3D6ytrbl6NYVt2/6utueKiGjDX39tIj8/H51Ox8aN643ar23b9qxb9xMAqanX2LlzO61btyM9PZ38/Hw6dOjE+PETcXBwIDHxMomJl7G3d6BXrz688MLLnDp10iwlRGRm9i6GdmtMdm4RP29PwNHOip5tAswdkhDiDoYOfZy3336dJ598DE9Pb9q0aVdtz9WlywMcPXqEkSMfx8nJidDQlmRllX/m2ksvvcoHH/yHkSMfR6/XM378RAIDG3Pq1Emio2eh1WrRarV07HgfoaEt2bAhhpUrv0apVKHX63jttTdQKk3/vV6h11fVFJRlqOwpsGWdoqjV6Zj3wzEOn73G2IGhdGjuXRVhVquadLpleWpLX2pSP65cOY+PT4MyH7eU4aaqUNm+5ObmYGdnj06n47//fQcPD0/Gjn2uGiI0njF9udN7axGnwNZkKqWS8YNC+WjlIRbGxGJvq6ZFI3dzhyWEMKN33pnGlSuJFBQUEBwcwvDhT5k7pGojScIIVhoVk4aEEf3NQeb9cIxXnwinsZ95zr4SQpjfe+/9n7lDMBmZuDaSnY2GyY+1wslewyerDpModZ6EEHWAJIkKcHaw5pWocFQqJR+uPETa9XxzhySEENVKkkQFebnaMfmxVuQXFvPhykNk5RaaOyQhhKg2kiQqob63I5MGh3E1I59PVh8hv7DY3CEJIUS1kCRRScH1XZkwKJSEK9eZ9+MxirW145RAIYS4lSSJexARVFLn6Xh8Wkmdp9p1yYkQZvfKK5P46ac1pe7T6/UMHTqIgwf3l7nfu+9O5/vvVwLw009rWLny6ztuFxPzM2+99a9y49iy5S9iY48Zbp88GcuMGW8Z0wWjDRkygLi4s1XaZlWQJHGPuob5MbRbY/acSOGb309Ty65NFMKs+vcfyPr1MaXuO3hwP0qlgvDw1ka18fDDQ4iKGn5PcWzd+hcnThw33G7WrDnTps26pzZrCrlOogo81LEBWblF/LrnAk52Vgzs0sjcIQlRZXYn7Wdn0l7D7aosFd7Jtx0dfNuU+XjXrg/w4YfvkZAQb1jL4ZdffqZfvwHExZ3jww//S35+HoWFhQwc+AiPPTbstjZuXU+iqKiIjz8uWS/C2dmF4OBmhu3OnTt7x/Z2797Jtm1b2LdvD+vWrSUqahje3j7Mm/epYQ2JDRti+Pbbr1AoFPj5BfCvf/0bV1e3O6xf4cCsWe/j7u5x19fl0qWLfPDBf8jISEelUjF27PN07Hgf+fn5zJo1jYSEOFQqNfXrN+Cdd/7LhQsJ/Oc/M8jLy0en0/LQQwMYNuzJyrwlt5EkUUWGdm9MVm4hP22Lx9FOQ/fWUudJiHul0Wjo3fsh1q//meeee5Hc3By2bv2bFStW4eDgwCeffIaVlRW5ubmMHTuS9u07GZLJnaxd+z1JSYmsWLGa4uJiJk4cg49PSYVbX1/fO7bXoUMnunS5n2bNQhg8OAqAAwf2GdqMizvLggVzWbRoBR4eHnz55Xw+/vgDZs58Dyi9fkV09Kwy16+41YwZbzFo0CNERj5MfHwcEyeOYcWKNRw5cojc3BxWrCipqH1zjYkfflhD164PMHz406XurwqSJKqIQqHg6X7NyM4rYsVvp7G31dA+xPLrPAlRng6+bUp92zd17ab+/Qfy6qsvMG7cRDZt+p2WLVvh5eVNWloqc+f+l7NnT6NQKLl27Spnz56+a5I4cGA/Dz0UiVqtRq1W07dvPw4dOghAfn5+hdsraXMfnTp1xsOj5Ohg0KBHefrp/x3R3Lp+RWhoC/bu3X3X9nJzczh79jT9+g0EoFGjQJo0Ceb48aM0adKUhIR4PvwwmoiINtx3XxcAwsMjmD9/Drm5ebRu3ZbWrduW86oaT+YkqpBKqWTCwy1oEuDMl+tiOR6fZu6QhKjxmjYNwt3dk127drB+/c/071/y4fn55/Nwc3Nn8eKvWbbsW0JCQiksrPx1S1Xd3k2l169Q3dO6FP7+AaxYsYp27Tqwb99unn76CQoKCujWrScLFiy68fhS3nln6j3HbYi5yloSQEmdpxeHhOHrbsfcH44Sl2ieJQeFqE369x/I4sVfcPHiBbp2fQCA7OwsvLy8UavVxMWd5fDhQ+W206ZNW379dT3FxcUUFOTz22+/Gh67W3v29vZkZ2ffsc3Wrduyc+d2UlOvAbBu3U+0a9e+0n21s7OnSZMgNmwombBPSIjn3LnThIa2JCUlGaVSxf33d2PSpFfIyEgnK+s6ly5dxN3dnX79BvDMM2OIjT1ezrMYT4abqoGdjYbJUeH856v9fLL6MG+MaI2vu725wxKixurduy/z5n3KwIGPoNFoABg58lneeWcqv/yylnr16hMeHlFuOwMHPsrZs2cZMWIozs4uhIQ0JzU1tdz2+vTpx7vvzmDz5k2GieubAgObMH78RF5++fkbE9f+vPbav++pv9OmzeKDD/7DqlXfoFKpeOutmbi6urJz53YWLJgLgE6nZcSIp/Hw8GT58sX8/vuvqNVqFAoFL774yj09/61kPYkbqqPef3J6Lu99tR+1Wsm/R7TBzcmmStsvS01au6A8taUvNakfsp5EzVRd60nIcFM18na14+XHwskrKKnzlJ1XZO6QhBCiQiRJVLMGPrfWeTosdZ6EEDWKJAkTCK7vyvhBocQnXeczqfMkaoBaNgotqPx7KknCRFoHeTKybzOOxaex6JcTUudJWKyS0zTliLe20WqLUSpVFd5Pzm4yoftb+ZGdV8Sav87hYKNhWO+mKBQKc4clRCm2tg5kZWXg4uKOQiHfI2sDvV5HVlY6trZlT1CXRZKEiT3UoT7Xcwr5be9FHO01DOwsdZ6EZXFwcCY9/SrJyZeA2494lUolOl3tGDKtO31RYGVlg4ODc4XblSRhYgqFgsd6NCE7r4iftsbjaGdF9wh/c4clhIFCocDNzavMx2vS6bzlkb6UT5KEGSgVCp5+6Eadp42ncLDV0K5Z2X+UQghhLiYbcIyPjycqKoo+ffoQFRVFQkLCHbdbv349AwYMIDIykgEDBnDt2jVThWhSalVJnafGAc588fNxjidInSchhOUxWZKYNm0aw4YNY+PGjQwbNoypU28vQHX06FHmzp3L4sWLiYmJ4ZtvvsHR0dFUIZqc9T/qPMUnSZ0nIYRlMUmSSE1NJTY2lsjISAAiIyOJjY0lLa30t+elS5cyatQoPD09AXB0dMTa2toUIZqNvY2Glx8Lx9FWw8erDpOUmmPukIQQwsAkcxJJSUl4e3ujUpWco6tSqfDy8iIpKQk3NzfDdufOnSMgIIDhw4eTm5tL7969mTBhQoVOE71bDZLyeHqa56jF09ORdyd05vW52/hkzRHen9gVDxfbe26ztqgtfakt/QDpi6Wqjr5Y1MS1Vqvl1KlTLFmyhMLCQkaPHo2fnx8PP/yw0W1YUoG/itAALw4JI/qbA7w5fztThrfGwVZTqbbM3ZeqVFv6Ulv6AdIXS1XZvlhEgT9fX1+Sk5MNi21otVpSUlLw9fUttZ2fnx99+/bFysoKBwcHevbsyZEjR0wRokW4WecpJT2XT1cfpqCw8ouTCCFEVTBJknB3dyckJISYmJJFNGJiYggJCSk11AQlcxXbtm1Dr9dTVFTErl27aNas2Z2arLWaNXBl3MAWxCVdZ95PR6XOkxDCrEx2dtP06dNZsWIFffr0YcWKFcyYMQOAMWPGcPToUQD69+9/Y3Wlfjz88MM0adKEIUOGmCpEi9Em+Eadp7g0FkudJyGEGcmiQzdY4tjkLzsT+P7vOHq1CeCJXsbXebLEvlRWbelLbekHSF8sVXXNSVjUxLUorV/HBmTlFt2o82TFgPsamjskIUQdI0nCgt2s85SVW8iPW+JwtNXQTeo8CSFMSJKEhVMqFDzTL4Sc/GK+ulHnqa3UeRJCmIgUi68BDHWe/J35Yt1xTkidJyGEiUiSqCGsNSpeHBqGt5sds384SsIVqfMkhKh+kiRqEHsbDZMfC8fBpqTO05W0XHOHJISo5SRJAEXaIk5fizN3GEZxdbTm1cfDAfjwu4OkZxWYOSIhRG0mSQI4mnqCtzZ9wOaL28wdilG83eyY/Fg4OfnFfLTyENl5ReYOSQhRS0mSAFp5hNLOvxVrzvzM7qT95g7HKA18HHlhcBjJ6bl8ukbqPAkhqockCUClVPFip2cJcm3CipOrOXL1uLlDMkpIA1fGDQwlLvE6n/10TOo8CSGqnCSJG6xUGsa1fIp6Dv4sOv41p9PPmTsko7QJ9uKpPsEcjUtl8Xqp8ySEqFqSJG5ho7bhuVaj8LBx4/MjS7lw/ZK5QzLKA+H+PHp/ILuOJ/PdpjPUsnJcQggzkiTxDw5W9kwMH42dxo55hxdxJSfF3CEZpX+nBvRqG8Af+y6xetMZc4cjhKglJEncgauNCy+Ej0aBgjmHviQtP93cIZVLoVDweM+mdAz15qsNJ/j70GVzhySEqAUkSZTBy86T58NHk19cwJxDX5JVmG3ukMqlVCgY1S+ENs28WL7xFPtP1YyjICGE5ZIkcRf1HP2Y0OoZ0vMzmXdoIXnFeeYOqVxqlZIpT7Uj0M+Jz38+zonzln8UJISwXJIkytHEpRGjW4zgcs4VFhxZSqHW8i9cs7FW8+KQVni72jHn+yOcv1I7FlURQpieJAkjtPAIYWRIFOcyElh8fAVaneVfuOZgq2FyVDj2Nho+WnWIZKnzJISoBEkSRmrrE8FjQQ9z9NoJvjqxGp3e8i9cc3W05pXHw9Hr4cOVh6TOkxCiwiRJVMD9AZ2IbNSHvckHWHNmXY24HsHHzY7JUa3Iyivio1WHyMm3/OEyIYTlkCRRQX0b9qBHva78fWk76xP+MHc4Rmno48SkR1uSnJbLp6uPUFBk+cNlQgjLIEmighQKBY806U8Hnzasj/+9xlSODWnoxtgBoZy7nMl8qfMkhDCSJIlKUCqUDG82hDCPUNac+Zk9Vw6YOySjtG3mxZN9gjlyLpUlUudJCGEESRKVpFKqGBU6jCCXxnx1YhVHr8WaOySjdIvw55Gujdh5PJlVf56tEfMqQgjzkSRxDzQqDePCRhLg4MfCYys4U0Mqx0be15BebQL4be9F1u86b+5whBAWTJLEPbJR2/B8q2dxt3FjwZGlXMiy/MqxCoWCx3s1pWNzb77/O44thxPNHZIQwkJJkqgCDlb2vBA+Glu1LfMOLSK5BlSOVSoUjOofQotAN5b9epL9p66aOyQhhAUyWZKIj48nKiqKPn36EBUVRUJCwm3bzJkzh06dOjFo0CAGDRrEjBkzTBXePXO1ceGFiDEAzDm0kPT8DDNHVD61SsnzD7ck0LekztNJqfMkhPgHkyWJadOmMWzYMDZu3MiwYcOYOnXqHbd7+OGHWbt2LWvXrmXatGmmCq9KeNt5MjF8NHnF+TWmcqy1lYoXh7bCy9WW2VLnSQjxDyZJEqmpqcTGxhIZGQlAZGQksbGxpKWlmeLpTaqeoz8TWj1DWn468w4vIq8439whlcvBVsPkx1phb6PmY6nzJIS4hdoUT5KUlIS3tzcqlQoAlUqFl5cXSUlJuLm5ldr2l19+Ydu2bXh6evLCCy8QERFRoedyd3eodJyeno6V3rd0O2FY24/lg20LWHziK/59/0Ss1FZV0rbxMVSsL56ejsya0JnX527j4zVHeH9iF3fIS5oAACAASURBVNydbaspuoqpqvfF3GpLP0D6Yqmqoy8mSRLGevzxxxk/fjwajYbt27fz3HPPsX79elxdXY1uIzU1G52u4uf+e3o6cvVq1Q211NM05MmQKJbFfkf03wsY0+IpVEpVlbV/N5Xti7UCXhwSxvvfHuSt+dt5fXhr7G001RCh8ar6fTGX2tIPkL5Yqsr2RalU3PXLtUmGm3x9fUlOTkarLakZpNVqSUlJwdfXt9R2np6eaDQlH0qdO3fG19eXM2dq7nrN7XwiGBo0iKPXTrDiZM2oHNvI14mJj7bkSloun66ROk9C1HUmSRLu7u6EhIQQExMDQExMDCEhIbcNNSUnJxt+PnHiBJcvX6ZRo0amCLHaPBBwH5GNHmTPlQN8X0Mqx4Y2dGPMgFDOXcpkgdR5EqJOM9lw0/Tp05kyZQqfffYZTk5OREdHAzBmzBgmTZpEy5Yt+eijjzh+/DhKpRKNRsP777+Pp6enqUKsNn0b9iSnOJfNF7dhr7GjX6Pe5g6pXO2aeZHdJ5ivNp5i6YaTjOofglKhMHdYQggTM1mSaNy4MatXr77t/i+//NLw883EUdsoFAoebRJJblEev8T/jp3Gjm4Bnc0dVrm6R/iTlVvIT1vjcbDVENWjCQpJFELUKRY1cV2b3awcm1ucx+rTa7FT29Lep7W5wyrXgPsakpVTxG97L+Jkb0W/jg3MHZIQwoSkLIcJqZQqng0dXqMqxyoUCp7o3ZQOzb1Z89c5qfMkRB0jScLENCoNY8NGEuDgy6JjKziTHmfukMqlVCh4tn8ILRqV1Hk6cFrqPAlRV0iSMANbtQ3PtXoWtxpUOVatUvL8Iy1p5OvEgrXHOXVB6jwJURdIkjATRyuHG5VjbUoqx+Za/rdzaysVLw1thaeLjdR5EqKOkCRhRqUqxx78skZUjnWw1fBKVDi21jfqPKVLnSchajNJEmbmbefJ8+HP3qgcu5Dswhxzh1QuNycbXokKR6eHD787REZ2gblDEkJUE0kSFqC+YwDjw54mLT+txlSO9XW356WhrcjKLeKjlYfJzS8yd0hCiGogScJCNHUN5NkWI7iUncjnR5ZSpLX8D91Av5I6T0mpOcxec4RCqfMkRK0jScKCtPRozpMhj3EmI45Fx79Gq7P8D93QRm6MGdCcM5cyWbD2OFqd1HkSojaRJGFh2vu0vlE5NpavT66pEZVj24d4M+LBIA6dvcbSDSdrRBFDIYRxjC7LsWvXLvz9/alXrx4pKSl8+OGHKJVKJk+eXCuK8FmSbgGdyS3KLanzpLZlcNMBFl8zqXvrAK7nFrF2WzyOtlY81qOJuUMSQlQBo48kZsyYYVhZLjo6muLiYhQKBW+//Xa1BVeXPdSwF90DurD50jZ+Tdhk7nCMMrBzQ3q09ufXPRfYsPu8ucMRQlQBo48kkpOT8fPzo7i4mG3btvHnn3+i0Wjo2rVrdcZXZykUCh5tGklOcS4x8b9hp7HjgYD7zB3WXSkUCob1DiI7r4jVm8/hYKuha5ifucMSQtwDo5OEg4MD165d48yZMzRu3Bh7e3sKCwspLi6uzvjqNKVCyYhmQ8krzmfV6Z+wU9vSzqdia36bmlKhYHRkc3Lyi1m64SQOthoimspwpBA1ldHDTSNGjGDIkCG8+uqrDB8+HIADBw4QGBhYbcGJ/1WObeoSyPITKzl27YS5QypXSZ2nFjT0cWL+T1LnSYiaTKGvwKko8fHxqFQq6tevb7hdWFhIcHBwtQVYUamp2eh0FT+7xtIXRM8rzmf2wc9JyklmYvgYmriUvayrpfQlK7eQ/359gIzsAl4f1pr63o4VbsNS+nKvaks/QPpiqSrbF6VSgbu7Q9mPV6SxRo0aGRLErl27uHr1qkUliNrsf5VjXZl/eAkXsy6bO6RyOdpZ8UpUODZWaj5adZgUqfMkRI1ToeGm/fv3A/DFF18wefJkXnnlFRYsWFBtwYnSSirHjsFWbcPcQwtrROVYQ50nnZ4PVx4iU+o8CVGjGJ0kzpw5Q3h4OACrV69m+fLlrFq1iu+++67aghO3c7Vx4YXw0UDNqRzr51FS5+l6ThEfrZI6T0LUJEYnCZ1Oh0Kh4MKFC+j1epo0aYKvry+ZmZnVGZ+4A297rxuVY/NqTOXYQD8nnn+0BYnXcpj9/VGp8yREDWF0kmjTpg0zZ84kOjqa3r17A3DhwgVcXV2rLThRtn9Wjs2vAZVjWzRyL6nzdDFD6jwJUUMYnSTee+89nJycCA4OZuLEiQDExcXx1FNPVVtw4u6auja+pXLsshpRObZ9iDfDepfUeVq24ZTUeRLCwhl9MZ2rqyuTJ08udV+3bt2qOh5RQTcrxy6L/Y4lx7/h2RYjzB1SuXq2CSArt5CftyfgaKdhaHep8ySEpTL6SKKoqIjZs2fTs2dPWrZsSc+ePZk9ezaFhYXVGZ8wQnuf1gxtOojD147zzcnva0Tl2EFdGtG9tT8bdl/g190XzB2OEKIMRh9JfPDBBxw5coQZM2bg5+dHYmIin332GdnZ2fz73/+uzhiFEbrV60xOcS7r43/nq0POPOT/oEVXjlUoFAzvFUR2bhGrNp/FwVZDlzBfc4clhPgHo5PEr7/+ytq1aw0T1YGBgTRv3pxBgwZJkrAQ/Rr2KikxfnoTymI1fRv2NHdId6VU3qzzVGSo8xTe1MPcYQkhbmH0cFNZE4zGTjzGx8cTFRVFnz59iIqKIiEhocxt4+LiaNWqFdHR0caGJyj5dj646QC6NmjPuriNbLm009whlUujVjLx0ZY08HFg/tpjnL5o+dd9CFGXGJ0k+vbty4QJE9i6dSvnzp1jy5YtPP/88zz00ENG7T9t2jSGDRvGxo0bGTZsGFOnTr3jdlqtlmnTptGrVy9jQxO3UCqUTGj/FC09Qlh1+if2XTlo7pDKZWOl5qWhrXB3suHTNUe4kFw7aukIURsYnSRee+01OnXqxMyZM3n00UeZNWsWHTp0QKPRlLtvamoqsbGxREZGAhAZGUlsbCxpaWm3bfvFF1/QrVs3GjZsaHwvRClqpYpRoSNo7NKQZTWkcuz/6jyp+HjVYVIy8swdkhCCClaB/aeCggLCw8M5ceLuH0LHjh3j9ddf55dffjHc169fPz744ANCQ0MN9508eZJ33nmH5cuX89lnn5Gbm8vrr79e2fDqvNzCPGb89TGXrl/h7Qcm0czT8k81vXDlOlPmbcPB1oroiV1wdbIxd0hC1GlGT1zfiUKhqLKLoYqKinj77bd57733DMukVkZtLRVeEbf2ZVzoM3x8YD7vbZnHixHjqedo2SvF2aoUTBocxgffHeTN+dt5f9L95GVb/tXk5amtv181nfSlikuF34kxp1n6+vqSnJyMVltSr0er1ZKSkoKv7/9Oebx69SoXLlxg7Nix9OjRg2XLlrFq1SpZQ/seOVo5MDF8NNYqa+YdWkhKDagc29jfmYmPtCTxWg4TP/iT/adS5MpsIcyk3COJnTvLPkOmqMi4MhDu7u6EhIQQExPDoEGDiImJISQkBDc3N8M2fn5+7N6923B7zpw5MtxURdxsXHkhfAwfH5jPnEMLmdx6Aq42LuYO665aBLozZXhrvv7jDPN+PEarxu4M7x2Eh4utuUMTok4pd06iR48e5Tby559/lrvNuXPnmDJlCtevX8fJyYno6GgCAwMZM2YMkyZNomXLlqW2r2ySkOGmsvty4folPj34OS7WzrzcegIOVvZmiK5iXN3s+XbDCX7aFgd6GNilEQ+2q4dadc8HwSZVF36/aiLpS/nDTfc0cW2JJEncvS9n0s8x9/Ai/O19mRQxBhu1ZU8M3+xLamY+3/xxmoNnruHvYc+TfYIJqmfZR0O3qiu/XzWN9MUEcxKiZmnq2phnQ4dzMfsynx9dXiMqxwK4O9vwwuAwXhjckrzCYv779QGWrD9Bdl7NiF+ImkqSRB0U5hnKiGZDOZ1+liXHv0GrqzkLAEU09WTW6A707VCf7Uev8O8vdrH9aJJMbAtRTSRJ1FEdfNswpOnAksqxp76vUR+yNlZqHuvehGnPtMPbzZZFv5zg/W8OknjN8lfoE6KmkSRRh3Wv14V+DXuxK2kfP5yNqVGJAqCelwNvjGjDyL7BXLqazbTFe/hhyzlZGlWIKnRPF9OJmq9fo97kFOfy58Wt2Gvs6duw/LPZLIlSoeCBcH8imnqy8s+zxOw4z+7YZEY8GEzLQHdzhydEjSdHEnWcQqFgSNOBtPOOYF3crzWicuydONlbMWZAc157IgKVUsnHqw4z/6djpGcVmDs0IWo0OZIQKBVKngx5jLzifFad/gk7jS1tvcPNHValhDRwZcao9mzYfZ6YHec5GpfKo/cH0qN1AEql5S7CJISlkiMJAYBKqeLZFjcqx8Z+x/HUk+YOqdI0aiUDOzfindHtaeznxDd/nOGd5ftIuHLd3KEJUeNIkhAGVioN48Oext/ehy+PfsW5jARzh3RPvF3tmBwVzvhBoWRkFfDOsn18/ftpcvOLzR2aEDWGJAlRiq3alufDR+Nq48z8I4u5lJVo7pDuiUKhoH2IN++O6UD3CH/+3H+JNxfuYu9JKRoohDEkSYjbOFo58EL4GKxV1sw9vJCU3GvmDume2dloGPFgMG+NbIuzvRXzfzrGx6tlcSMhyiNJQtzRzcqxer2euYe+JKMg09whVYlGvk68PbItT/RsyplLmby9cDcxOxIo1urMHZoQFkmShCiTj70Xz7d6lpyiXOYcWkh2Ue24olmlVNK7XT3+M6YjYY3d+WFLHNMW7+HUhXRzhyaExZEkIe6qvlMA48Ke5lpeKp8dXkx+cc1fJe4mV0drnn+kJS8OCaOoWEf0NwdZFBPL9dxCc4cmhMWQJCHKFeTamFGhw7mYdZkvji6nSFe7zg5q1cSDd0Z3oH+nBuyKTebNL3ax5XAiOpnYFkKShDBOqxuVY0+ln2VpDascawxrjYrBDzRm+jPt8PewZ+mGk0R/fYBLV7PNHZoQZiVJQhjtZuXYQ1eP8e2pH2rlKaT+ng68Prw1z/RrRlJqLjOW7GX1X2cpKKxdSVEIY0lZDlEh3et1Iacolw0Jf2CntuWRJv1RKGpXuQuFQkHXMD/Cm3iw+q9zbNh1gT2xKQx/MIjwJh7mDk8Ik5IjCVFh/Rv15oGA+9h0cQu/nd9s7nCqjaOdFaP6hfD6sAisrVTMXnOEuT8cJe167Zm8F6I8kiREhd2sHNvWO5yf435l6+Vd5g6pWgXXd2X6M+0Y/EAgx+JSeXPhbn7bcwGtTq6tELWfDDeJSlEqlDwVEkV+cT4rT/2IndqGNjW0cqwx1Col/Ts1pH2INyt+O813f55l+7ErPNU3mMZ+zuYOT4hqI0cSotJKKsc+SaBzQ5bFruR46ilzh1TtPF1seWloGM893IKs3EL+s3w/X208RW5+kblDE6JaSJIQ98RKpWFCq6fxtffmy6PLa3zlWGMoFAraNvPi3TEd6dk2gL8OXebfX+5m1/ErtfKML1G3SZIQ98xWbctEQ+XYJVzOTjJ3SCZha61mWK8gpo5sh5ujNV+si+XDlYdITss1d2hCVBlJEqJKOFo5MLHVGKxVVsw59GWtqBxrrAY+jrz1VFuG9w4iPuk6by/aw9pt8RQVy8S2qPkkSYgq427rygvho9HpdbWqcqwxlEoFPdsE8O6YjrQO8mDttnimLt5DbEKauUMT4p5IkhBVysfem+dbPUt2UQ5zDy0kp6huDb24OFgzflALJj/WCr1Oz/99d4gv1h0nPUuurRA1k8mSRHx8PFFRUfTp04eoqCgSEhJu2+b7779nwIABDBo0iAEDBrB8+XJThSeqUAOneowPe5qrhsqxBeYOyeRaBLoz89n2DLivIXtPpDAh+k/+OnhZigaKGkehN9HpGE899RSDBw9m0KBBrF27lu+///62JJCdnY29vT0KhYLs7GwGDBjA/PnzadasmdHPk5qajU5X8S55ejpy9WpWhfezRJbSl8NXj7Pw2FcEuTRmfKtn0CgrflmOpfTlXiSl5rBy8zmOnL1GYz8nnuwTTH1vR3OHVWm14T25SfpSMlTq7u5Q9uP3EpSxUlNTiY2NJTIyEoDIyEhiY2NJSys9Xuvg4GCoA5Sfn09RUVGtqwtUl7TyDGV4syGcTD/D0uPfotPXzYlcX3d7Zo2/jzGRzUnJyGPm0n2s/PMM+YW1q+S6qJ1McsV1UlIS3t7eqFQqAFQqFV5eXiQlJeHm5lZq202bNvHRRx9x4cIFXnnlFYKDgyv0XHfLiOXx9Ky53+7+yVL6MsCzO0prHcsOreHHhJ8Z125EhRO/pfTlXg3s3pTuHRqw7JdYNu46z/5TVxn7SBidWvqaO7QKqy3vCUhfymNxZTl69uxJz549SUxM5Pnnn+f+++8nMDDQ6P1luMny+tLerT0pDdPZEL8JpVbDI036G72vpfWlsm7tR1S3xrRp4sHyjSf5z9I9hDfxYFjvpng425o5SuPUlvcEpC9gIcNNvr6+JCcno9WW1OTXarWkpKTg61v2Nyg/Pz9atmzJX3/9ZYoQRTXr3+hB7ve/jz8u/F2rK8caq0mAM1Ofbsdj3ZsQez6NtxbuZsPu8xRr6+aQnLBcJkkS7u7uhISEEBMTA0BMTAwhISG3DTWdO3fO8HNaWhq7d+8mKCjIFCGKaqZQKBgaVFI5du25DWyr5ZVjjaFWKenboT7vju5I8wZurN58jplL93L2Ut25vkRYPpMNN02fPp0pU6bw2Wef4eTkRHR0NABjxoxh0qRJtGzZkpUrV7J9+3bUajV6vZ4RI0bQpUsXU4UoqtmtlWO/O/Ujtmpb2ni3MndYZufubMOkIWEcPH2Vr/84zX9W7Of+Vn4M6dYYB1uNucMTdZzJToE1FZmTsPy+FGoLmXtoIQnXLzI+7Gmau5d9coKl98VYxvYjv7CYn7cl8Nvei9jbqonq0YROoT4WdZZfbXlPQPoCFjInIcStrFRWjA97xlA5Ni4zwdwhWQwbKzWP9WjC1Kfb4uViy8KYE3zw7UGSUnPMHZqooyRJCLOw09jyfPizOFs78dnhulM51lj1vR1548k2PNU3mAvJ2UxdtIcftsRRWKQ1d2iijpEkIczGycqRF8JLKsfOPbSQq7mp5g7JoigVCrqF+/Pu2I60D/EiZkcCUxft4Vi8vE7CdCRJCLNyt3VjYvhotHotc+pY5VhjOdtbMWZAKK89Ho5CqeCjlYdZsPYYGdl1ryaWMD1JEsLsfA2VY7OZd2hRnasca6yQhm7MHNWeh7s04sDpa7z55S427b9UqRM1hDCWJAlhERo41WNcy6dJyb3K/DpaOdYYGrWSgV0a8c6z7Wnk68TXv59m1vJ9nL9SO87QEZZHkoSwGMFuTRjVYjgJ1y/y5dHlFOmkAF5ZvN3seCUqnLEDm5OWVcDMZXv55vfT5BXIa1YXFGqLuJKTQmzqKbZd3sVv5zdzvSC7Wp7L4mo3ibqtlWcLhocMZcWJVSw7/i2veY4zd0gWS6FQ0LG5D2GB7ny/JY5N+y+x71QKw3oF0SbY06KurRAVk1uUR1p++o1/Gbf9nFVUOiFolGo6NArDGfcqj0UuprtBLqqxLH9e2ML3Z2NQKBQ4WznhbO2Ei7XzjX+3/lzyz0pl2Vcmm+I9iUu8zvJfT3IhJZuWge4MfzAIL5eqLxpYG36/bjJHX/R6PVlF2XdIACW3U/PSydeWXslQo1TjZuN645/LLT+X3Ha2csLH26VaLqaTJHGD/OJbnsNXj3NNm0Ji+lUy8jPJKLxOZkEmecW3LwVqp7b9R+JwwsWmdCKxU9ua7du1qd4TrU7Hpv2X+XFrHDqdnoGdG9KnfX3UqqobWa4tv19QPX3R6rRkFl6/LQGk5qWTVpBOen7GbUOptmqbMhOAm40rjhqHcn93q+uKaxluEharlWconp4db/vFzy8uILMgk/SCTDILrt/4P9Pw/6XsRLIKs9FT+suCRqm+7QjE2doJV2tnnK2dcbVxxlHjgEqpMmU3q5RKqeTBdvVoG+zJt5vO8P3fcew4doWn+gQTXN/V3OHVCkXaItILMgxJIPUfRwIZBZm3LbDlqHHAzcYVf3tfWro3NyQAd1s33GxcsFVbbpl4SRKixrFRW2Oj9sLb3qvMbW5+m8soyCSj4DoZ+Rkl/xdkklGQSVzmeTILMinWl76CWYECZ+uaP7zl5mTD84+05PDZa3z9+2mivzlIl5a+DO3eGEc7K3OHZ9Hyi/NLhn3y0+44H3C9sPSXFgUKXKydcbNxobFzw/8lAJuSBOBq42rxvy93I0lC1EoqpcpwyF4WvV5PdlGOIXEYEkpBJhn5mSTnXuVU2tnbxocB7NV2tycSCxreuqlVEw+aNXBl3fYENu65wMEzV3msexM6h/mirIMT2zff85sf+oVpuVy4dqVUMsgtziu1j1qhwvXGsE+oe7NSCcDNxhUXa+caffRZHkkSos5SKBQ4WjngaOVAPUf/MrfLL84vdRSSUXC91PDWxezLZBfm3GF4S2M4CvF29sBWb2eW4S1rjYoh3RrTKdSb5RtPsWTDSbYfTeLJPsH4e1Z+uV9LpNPryCy4fT7g1tuFuqJS+1irrAxfKBo5N8D9H3MDjlYOKBV192oBmbi+QSbjLFNN6UuxrpjMgiwyCzNJzy9JHrcmluvFWaTnZph9eEun17P9SBKrNp8lv1BLn/b1GdC5IdYa4xOVOd+TYl0x6fmZZSaA9IJMtP94je01drjZuN748C89IRzkX4/cTK3Zj/iqgkxcC2HB1Eo17rauuNu6gvPtj3t6OpKckklOUe7tw1v5JT8n56RU+/CWUqGgays/wpt6sGrzWdbvOs+eE8mMeDCIsMYeVfVyVFqBtvCOCSA1L90wH3DrEZsCBU5WjrjbutLQuT6t/3EU4Grtgo3ausznc7C2J09h+V9CzEmShBAmolQoKz28dWtiuZh9mazC26+uvXV465+nA98c3nKyciyJw86KZ/s3p0tLX5ZvPMUnq4/QJtiTJ3o2xc3Jplr6r9fryS2+/SKxW88O+mfdLqVCiZt1yYd+iFvQbaeIuto4o1bKx1h1kldXCAtjo7bBR22Dz13O3rrT8NatpwTHZSaQUXD9tqGXW4e3bs6NdO3lyIVLxRw4fppjyy8yqH0Ivds2QKWs2Di8Tq8jqzDb8IGfeoeLxQq0haX2sVJqDB/4DRwDbr9IzNqpTs8HWAJJEkLUQOUNb0HJh/Ztw1v5/zsyuZKTwsm0M+RrS4opqoJK9luX/Rcxm6xwt3PBy971tuEtD50T8cmJt88H5N8+52KntsXNxhVPWw+CXZuUSgDuNm7Ya+xqxXxAbSZJQohaqjLDW+n5mcReTuTQ+UskX88hz+UaF6wukV105+VTnawccbNxpZ6jP608W5RKAK42Ltiqq2foSpiOJAkh6rh/Dm918oO8sGJ+3BLHpn2XcLKzYniPQIIDbcgouI69kwZlvhWu1i5oavBFYsI4MtgnhLiNrbWaYb2DeHtkW1wdrVm47iRLf76Ag96Llt7N8LLzlARRR0iSEEKUqaGPE2891ZbhvYOIS8zk7YV7WLLuOCfOp1NQpC2/AVHjyXCTEOKulEoFPdsE0DrIk5V/nuHHv8+i14NKqaChjyNN67kQFOBCkwBnHGzl6KK2kSQhhDCKq6M14we14OXhNuw+fInTFzM5fSmDP/Zd5NfdFwDw97QnKMCFpvWcCQpwqbZrLoTpSJIQQlSIg62GsMYehiu0C4u0xCdd5/SlTE5fzGDH8StsPngZAA9nG4LquRBUz4WmAc74uMkprzWNJAkhxD2x0qgIru9qWK9Cq9NxMSWb0xczOXMxg6Nxqew4dgUAJzsNTQNcSoao6jlTz8uhwhftCdMyWZKIj49nypQpZGRk4OLiQnR0NA0bNiy1zbx581i/fj1KpRKNRsPLL79M165dTRWiEKIKqJRKGvo40dDHiQfb1UOv13MlLZczN440Tl/MYP/pqwDYWKlo7O9MUIAzQfVcaOTrhFUFig2K6meyJDFt2jSGDRvGoEGDWLt2LVOnTmX58uWltgkLC2PUqFHY2tpy8uRJRowYwbZt27CxkXFNIWoqhUKBr7s9vu723N/KD4C06/n/SxqXMvhxazwAapWChr5OBAWUHGk08XfGzkYmw83JJEkiNTWV2NhYlixZAkBkZCTvvPMOaWlpuLm5Gba79aghODgYvV5PRkYGPj4+pghTCGEibk42dGhuQ4fm3gBk5xVx9lLJRPiZixls3HOB9bv0KIAAL4f/TYbXc8HFoeyqrqLqmSRJJCUl4e3tjUpVchipUqnw8vIiKSmpVJK41U8//UT9+vUrnCDuVhe9PJ6ejpXe19JIXyxPbekHVH1fPIFG9d3ofeN2fmExpy+kczwujdi4VLYfS2LTgUsA+Lrb0zzQjdBG7oQGuuPrYX9Pk+HyvtydRU5c79mzh08//ZTFixdXeF9ZdEj6YolqSz/AdH3xdbbBN8KPXhF+FGtvToaXzGnsPnaFTXsvAuBsb3XjWo2SI40ATweUSuOShrwvFrLokK+vL8nJyWi1WlQqFVqtlpSUFHx9fW/b9uDBg7z22mt89tlnBAYGmiI8IYSFU6uUNPJ1opGvE33a10en15OUmsuZG3MaZy5msO9kCgC21iqa+JfMaTQNKJkM16jlDKrKMkmScHd3JyQkhJiYGAYNGkRMTAwhISG3DTUdOXKEl19+mdmzZxMaGmqK0IQQNZBSocDfwx5/D3u6RZRUuE3NzDckjNOXMvn+71SgJMEE+t64MryeC038nbG1tshBFItksjWuz507x5QpU7h+/TpOTk5ER0cTGBjImDFjmDRpEi1btmTw4MFcvnwZb29vw37vv/8+wcHBRj+PDDdJXyxRbekH1Jy+ZOUWGibDT1/M4PyVbHR6PQoF1PdypGk9Z9qG+uLtZI2zvZW5zv3FqgAAC+5JREFUw71n1TXcZLIkYSqSJKQvlqi29ANqbl/yC4s5l3i95EjjYgZxidcpLNYB4O1mZ5jTaFrPBU9nmxp3ZXiNnpMQQghzs7FSE9rQjdCGJcPcxVodmQVa9hxN5MzFTA6cvsrWI0kAuDhY3SglUjJE5e9pj7KGJY2qIklCCFEnqVVKmjVwxt1Ow0MdQKfXk3gtxzCncfpiBntOlEyG29uoaeL/vyONhj6OqFV1YzJckoQQQlAyGR7g6UCApwPdWweg1+u5lpnP6YsZnLmUwemLmRw+VzIZbqVWEujnZDjSaOzvhI1V7fw4rZ29EkKIe6RQKPB0scXTxZbOLUtO17+eU2hIGKcvZRCzMwH9jpIEU9/boVTFW0e7mj8ZDpIkhBDCaE72VrQJ9qJNcMl64HkFxZxLzDRUvP3zwGV+u3GRn6+7XUnSuFFSxMPZ1pyhV5okCSGEqCRbazUtGrnTopE7AEXFOhKuXL8xRJXJnhMp/H0oEQA3J+sbCaPk6nBfj5oxGS5JQgghqohGrSxZLyPABQCdTs+lq9mGircnLqSzKzYZKFm86eZkeFA9F+p7O1jkZLgkCSGEqCZKpYL63o7U93akZ5uSyfCrGXmGOY3TFzM4dPYaAFYaJY39biSNAGcC/Z2xtoC1NSRJCCGEiSgUCrxc7fBytaNLWMlkeGZ2geGU2zMXM/h5Wzx6QKVU0MDH0TCn0TTABQdb06+tIUlCCCHMyNnBmnbNvGjXrGQyPDe/mLOXM2+cRZXBH/sv8uueCwD4e9gbln4NCnDBzan6F2STJCGEEBbEzkZNWGN3whrfnAzXEp+UZSiTvuv4Ff46eBkAD2cbmga40Ky+C4O6N62WeCRJCCGEBdOoVYbJbQCtTsellBzD0q/H41PZefwKDeu5Us+t6k+zlSQhhBA1iEqppIGPIw18HOndrh56vZ7svCICG7hXS+FFyzvfSgghhNEUCkW1Xt0tSUIIIUSZJEkIIYQokyQJIYQQZZIkIYQQokySJIQQQpRJkoQQQogy1brrJJTKypfevZd9LY30xfLUln6A9MVSVaYv5e2j0Ov1+soGJIQQonaT4SYhhBBlkiQhhBCiTJIkhBBClEmShBBCiDJJkhBCCFEmSRJCCCHKJElCCCFEmSRJCCGEKJMkCSGEEGWqdWU57iY6OpqNGzdy+fJl1q1bR1BQ0G3baLVaZs2axdatW1EoFIwdO5ahQ4eaIdq7M6Yvc+bM4ZtvvsHLywuA1q1bM23aNFOHelfp6en861//4sKFC1hZWdGgQQNmzpyJm5tbqe3y8vJ44403OH78OCqVitdff53u3bubKeo7M7YvU6ZMYceOHbi6ugLQt29fJkyYYI6Qy/Tcc89x6dIllEoldnZ2vP3224SEhJTapqb8rRjTl5rwt3KruXPnMmfOnDv+7Vf534q+Dtm7d68+MTFR3717d/2pU6fuuM2PP/6oHzVqlF6r1epTU/+/vbsNaaoN4wD+38llhblhLd3SUiMqqGi4sjSL5kujQE2y/CDEWsbIzF4MC6EXszA/jDRGZoVQH8QUsQ+6SBNJLSMpSqOFWK5w02FKMZOMPM+Hhw7tmUdnj20urx8MdnZu3HU5Ly7Pve2+P7FRUVHsx48fXRzpxJzJpaioiM3Pz3dxZJMzODjItra2csf5+fns6dOnHcZdvXqVzcnJYVmWZd+/f89GRESwNpvNZXE6w9lcsrOz2Tt37rgytEn78uULd7+uro5NTEx0GOMpteJMLp5QKz91dHSwGo2Gt/anulZm1HSTQqGAVCodd0xtbS2Sk5PBMAz8/PwQExOD+/fvuyhC5zmTiycQi8UIDw/njtetWwez2ewwzmAwYO/evQCA4OBgrF69Go8ePXJZnM5wNhdPMH/+fO6+zWaDQOC4CJyn1IozuXiKkZER5Obm4ty5c7xjprpWZtR0kzMsFgtkMhl3LJVK0dvb68aI/p+amho0NzdDIpEgIyMDcrnc3SHxGh0dRVlZGZRKpcM5s9mMxYsXc8fT/XUZLxcAKC0tRXl5OYKCgnDixAksW7bMxRFOLCcnBy0tLWBZFjdv3nQ470m1MlEugGfUSmFhIeLj4xEYGMg7ZqprhZrEXywlJQVarRZCoRAtLS04dOgQamtrubnw6ebChQuYN28eUlNT3R3K/zZeLseOHYNEIgHDMKiursaBAwdQX1+PWbNmuSFSfhcvXgQAVFdXo6CgADdu3HBzRL9volw8oVZevHiBjo4OZGVlufR5Z9R0kzOkUqndFIHFYkFAQIAbI/p9EokEQqEQABAZGQmpVIrOzk43RzW2y5cvw2Qy4cqVK2AYxz9LmUyGnp4e7ng6vy4T5eLv7889npiYiK9fv07b/8CBf2N8+vQpBgcH7R73xFrhy8UTauXZs2fo6upCdHQ0lEolent7odFo0NzcbDduqmuFmsR/qFQqVFRUYHR0FAMDA6ivr8f27dvdHdZv6evr4+6/efMGPT09CAkJcWNEY9PpdOjo6IBer8fs2bPHHKNSqVBeXg4A6O7uRnt7O6KiolwZplOcyeXX16WpqQkMw8Df399VIU5oaGgIFouFO25oaIBIJIJYLLYb5wm14mwunlArBw8eRHNzMxoaGtDQ0ICAgADcunULmzdvths31bUyo6ab8vLy8ODBA/T390OtVkMsFqOmpgZpaWk4cuQI1qxZg4SEBLx8+RJxcXEAgPT0dAQFBbk5ckfO5KLT6fD69WswDAOhUIiCggJIJBJ3h26ns7MT169fR3BwMFJSUgAAgYGB0Ov1SEhIQElJCfz9/aHRaHDq1CnExsaCYRjk5ubCx8fHzdHbczaX7OxsfPr0CQKBAD4+Prh27Rq8vKZPKQ4PDyMzMxPDw8NgGAYikQjFxcUQCAQeVyvO5uIJtTKeP1krtDMdIYQQXjTdRAghhBc1CUIIIbyoSRBCCOFFTYIQQggvahKEEEJ4UZMgZBpYsWIFTCaTu8MgxMH0+XA2IdOIUqlEf3+/3VIZu3btwpkzZ9wYFSGuR02CEB7FxcWIiIhwdxiEuBVNNxEyCVVVVUhJSUFubi7CwsKgUqnw5MkT7nxfXx+0Wi02bNiA2NhY3L17lzv348cPFBcXIyYmBnK5HElJSXZLRjx+/BhxcXFQKBQ4f/48fn7P1WQyITU1FWFhYQgPD8fRo0ddlzCZ8ehKgpBJevXqFVQqFVpbW1FXV4fDhw/j4cOHEIvFOH78OJYvX46mpia8e/cOarUaQUFB2LRpE0pLS1FTU4OSkhKEhITg7du3mDNnDvdzGxsbUVlZCZvNhqSkJGzbtg1btmxBYWEhIiMjcfv2bXz//h3t7e1uzJ7MNHQlQQiP9PR0KBQK7vbzqsDPzw/79u2DUCjEjh07EBISgsbGRlgsFjx//hxZWVnw9vbGqlWrkJycjHv37gEAKioqkJmZidDQUAgEAqxcudJuKeq0tDT4+vpCJpMhPDwcRqMRAODl5QWz2Qyr1Qpvb28oFArX/zLIjEVNghAeer0ebW1t3G3Pnj0A/l3q+9fdzWQyGaxWK6xWK0Qikd1iajKZjFthtLe3F0uWLOF9vl8XlJs7dy6GhoYAACdPngTLsti9ezd27tyJysrKKc2TkPHQdBMhk9TX1weWZblGYbFYoFQqsWjRInz+/Bk2m41rFBaLhVsGPCAgAB8+fHDYuH4iEokEeXl5AIC2tjao1WqsX78eS5cuncKsCBkbXUkQMkkDAwPc+wMGgwFdXV3YunUrpFIp5HI5dDodvn37BqPRiMrKSsTHxwMAkpOTUVhYiO7ubrAsC6PR6LD5zVgMBgO3KZFIJIJAIBhzMyNC/gS6kiCEh1artfueREREBKKjo7F27VqYTCZs3LgRCxcuRFFREffegk6nw9mzZxEVFQVfX19kZGRwH6NVq9UYGRnB/v37MTg4iNDQUOj1+gnjaG9vx6VLl2Cz2bBgwQLk5ORMu30byN+L9pMgZBKqqqpQUVGBsrIyd4dCiEvQNSshhBBe1CQIIYTwoukmQgghvOhKghBCCC9qEoQQQnhRkyCEEMKLmgQhhBBe1CQIIYTwoiZBCCGE1z9BHR3PeBS5JAAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AZ07lwVCtYGl"
      },
      "source": [
        "#save model\n",
        "model.save('model.h5')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BW9uaWZlZfgB",
        "outputId": "d71ab632-9859-4c51-b76b-9ae7dc6eee02",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "def feedback_learing(text,correct_pred):\n",
        "  X=clean_text(text)\n",
        "  tokenizer.fit_on_texts(X)\n",
        "  X = tokenizer.texts_to_sequences(X)\n",
        "  X = pad_sequences(X, padding='post', maxlen=300)\n",
        "  Y = np.asarray(correct_pred).astype('float32').reshape((-1,1))\n",
        "  print(model.train_on_batch(X,Y))\n",
        "  model.save('model.h5')\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[0.26442912220954895, 0.9006973505020142]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rbi6l6GmN74w"
      },
      "source": [
        "def predict_sentiment(input):\n",
        "  input=clean_text(input)\n",
        "  model_input=tokenizer.texts_to_sequences(input)\n",
        "  model_input=pad_sequences(model_input, padding='post', maxlen=300)\n",
        "  pred = model.predict(model_input, verbose=True)\n",
        "  pred=pred.reshape(pred.shape[0],)[0]\n",
        "  if(pred>0.5): return \"Satisfied\",pred\n",
        "  return \"Unsatisfied\",pred\n",
        " "
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GLpQ_-hnSRBW",
        "outputId": "166c751f-0a01-40e7-9288-a401cb6a710d",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "source": [
        "if __name__ == \"__main__\":\n",
        "  while(True):\n",
        "    user_messege=input(\"Hi, please enter a message to test model prediction: \")\n",
        "    pred,score=predict_sentiment(user_messege)\n",
        "    print(pred,'\\n',\"score: \",score)\n",
        "    feedback=input(\"Was it the correct prediction? [Y/N]\")\n",
        "    if(feedback=='N'):\n",
        "      if(pred==\"Satisfied\"): feedback_learing(user_messege,0) \n",
        "      else:  feedback_learing(user_messege,1)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "7/7 [==============================] - 1s 126ms/step\n",
            "Unsatisfied \n",
            " score:  0.40435666\n",
            "1/1 [==============================] - 0s 2ms/step\n",
            "Unsatisfied \n",
            " score:  0.4669529\n",
            "1/1 [==============================] - 0s 2ms/step\n",
            "Unsatisfied \n",
            " score:  0.40435666\n",
            "1/1 [==============================] - 0s 1ms/step\n",
            "Satisfied \n",
            " score:  0.5999802\n",
            "1/1 [==============================] - 0s 2ms/step\n",
            "Satisfied \n",
            " score:  0.64582694\n",
            "2/2 [==============================] - 0s 46ms/step\n",
            "Satisfied \n",
            " score:  0.5999802\n",
            "1/1 [==============================] - 0s 1ms/step\n",
            "Satisfied \n",
            " score:  0.5342141\n",
            "[0.6682041883468628, 0.6363636255264282]\n",
            "1/1 [==============================] - 0s 2ms/step\n",
            "Unsatisfied \n",
            " score:  0.47690475\n"
          ],
          "name": "stdout"
        }
      ]
    }
  ]
}