NLP Emoji Prediction Using Deep learning.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<H2><CENTER> Emoji Prediction<CENTER><H2>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Importing necessary packages"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import re\n",
    "import codecs \n",
    "import numpy as np  \n",
    "import pandas as pd  \n",
    "import time  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import spacy#package to import stop list\n",
    "from spacy.lang.te import STOP_WORDS as STOP_WORDS_TE\n",
    "with codecs.open(\"Data/telugu_stopwords.txt\", encoding='utf-8') as f:\n",
    "    stop_txt_te = f.read()\n",
    "stop_te = stop_txt_te.splitlines()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split, GridSearchCV#to split the data into test and train\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from sklearn.metrics import accuracy_score, f1_score, precision_score, recall_score, confusion_matrix\n",
    "#importing classifer algorithm\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "from sklearn.naive_bayes import GaussianNB\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.feature_extraction.text import CountVectorizer, TfidfVectorizer "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Reading files"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "24ఎస్టీయూ తెలంగాణ24 చరిత్రలో ఈ రోజు/అక్టోబర్ 24 󾔖ఐక్యరాజ్య సమితి పతాకం󾔖 󾔖ఐక్యరాజ్యసమితి దినోత్సవం󾔖 󾭺1577 :...\n",
      "ఈరోజు పుట్టిన రోజు జరుపుకుంటున్న మా అభిమాన హీరోకి ఇవే మా హృదయపూర్వక జన్మదిన శుభాకాంక్షలు.. #ప్రభాస్\n",
      "ఈరోజు పుట్టిన రోజు జరుపుకుంటున్న మా అభిమాన హీరో #ప్రభాస్ కిఇవే మా హృదయపూర్వక జన్మదిన శుభాకాంక్షలు\n",
      "*_BE A LIGHT INTO YOURSELF_* *మనం మానసిక ప్రశాంతతను ఎలా పొందవచ్చు.* మానసిక ప్రశాంతత ఇది ఎక్కడ...\n",
      "అమ్మకు ఎంత గౌరవమిస్తున్నారో భార్యకు కూడా అంతే ఇవ్వండి.. ఎందుకంటే ఒకరు *చందమామ* చూపిస్తారు.. మరొకరు *చుక్కలు* చూపిస్తారు..!!\n",
      "నీ ప్రమేయం లేకుండా నిన్ను ఎవరూ నాశనం చేయలేరు, అది తెలుసుకుంటే.. నీకు ఒకర్ని నిందించే అవసరం ఎప్పుడూ రాదు; శుభోదయం..!!\n",
      "పరిస్థితులెలా ఉన్నా నీ కర్తవ్యం నీకు గుర్తుంటే.. జరగాల్సిన పనులు జరుగు తుంటాయి; శుభోదయం..!!\n",
      "నీకు అందని దానికోసం ఆశపడకు, నీకు నచ్చని దానికోసం కష్టపడకు, నీకై కష్టపడేవారిని మరిచిపోకు, నిన్ను ఇష్టపడేవారిని వదులు కోకు. శుభోదయం..!!\n",
      "దీపావళి శుభాకాంక్షలు. మన పద్దతి ప్రకారమే అయితే దీపాలు ఆర్పకూడదు. అలా ఈసారి దీపావళి నాకు ఓరోజు ముందే వచ్సినట్టే.\n",
      "krish tweet ....\n"
     ]
    }
   ],
   "source": [
    "with codecs.open(\"Data/telugu.txt\", encoding='utf-8') as f:\n",
    "    txt1 = f.read()#to read the total file\n",
    "print(txt1[0:1000])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "emoji_te = (pd.read_csv(\"Data/telugu.csv\"))\n",
    "emoji_te = emoji_te[\"label\"].tolist() "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Cleaning Text"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "def remove_punct_te(text):\n",
    "    new_words = \" \"\n",
    "    for word in text:\n",
    "        if word not in STOP_WORDS_TE:\n",
    "             if word not in stop_te:\n",
    "                    w = re.sub(r'[.#-:?!\\_|\"\"'',-...A-Za-z0-9]','',word) #remove everything except words and space#how \n",
    "                    new_words = new_words + w\n",
    "    return re.sub(r\"\\s+\", \" \", new_words, flags=re.UNICODE)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[' ఎస్టీయూ తెలంగాణ చరిత్రలో ఈ రోజుఅక్టోబర్ \\U000fe516ఐక్యరాజ్య సమితి పతాకం\\U000fe516 \\U000fe516ఐక్యరాజ్యసమితి దినోత్సవం\\U000fe516 \\U000feb7a ', ' ఈరోజు పుట్టిన రోజు జరుపుకుంటున్న మా అభిమాన హీరోకి ఇవే మా హృదయపూర్వక జన్మదిన శుభాకాంక్షలు ప్రభాస్', ' ఈరోజు పుట్టిన రోజు జరుపుకుంటున్న మా అభిమాన హీరో ప్రభాస్ కిఇవే మా హృదయపూర్వక జన్మదిన శుభాకాంక్షలు', ' మనం మానసిక ప్రశాంతతను ఎలా పొందవచ్చు మానసిక ప్రశాంతత ఇది ఎక్కడ', ' అమ్మకు ఎంత గౌరవమిస్తున్నారో భార్యకు కూడా అంతే ఇవ్వండి ఎందుకంటే ఒకరు చందమామ చూపిస్తారు మరొకరు చుక్కలు చూపిస్తారు', ' నీ ప్రమేయం లేకుండా నిన్ను ఎవరూ నాశనం చేయలేరు అది తెలుసుకుంటే నీకు ఒకర్ని నిందించే అవసరం ఎప్పుడూ రాదు; శుభోదయం', ' పరిస్థితులెలా ఉన్నా నీ కర్తవ్యం నీకు గుర్తుంటే జరగాల్సిన పనులు జరుగు తుంటాయి; శుభోదయం', ' నీకు అందని దానికోసం ఆశపడకు నీకు నచ్చని దానికోసం కష్టపడకు నీకై కష్టపడేవారిని మరిచిపోకు నిన్ను ఇష్టపడేవారిని వదులు కోకు శుభోదయం', ' దీపావళి శుభాకాంక్షలు మన పద్దతి ప్రకారమే అయితే దీపాలు ఆర్పకూడదు అలా ఈసారి దీపావళి నాకు ఓరోజు ముందే వచ్సినట్టే', ' దీపావళి శుభాకాంక్షలు మన పద్దతి ప్రకారమే అయితే దీపాలు ఆర్పకూడదు అలా ఈసారి దీపావళి నాకు ఓరోజు ముందే వచ్సినట్టే']\n",
      "Time taken to filter the sequence : 7.794719219207764 s\n"
     ]
    }
   ],
   "source": [
    "start=time.time()\n",
    "text_te = txt1.splitlines()\n",
    "#print(remove_punct(text_cleaned[0:10000]))\n",
    "filtered_sentence_te = []\n",
    "for i in range(0,len(text_te)):\n",
    "    filtered_sentence_te.append(remove_punct_te(text_te[i]))\n",
    "print(filtered_sentence_te[0:10])\n",
    "end=time.time()\n",
    "tt=end-start\n",
    "print(\"Time taken to filter the sequence :\",tt,\"s\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Text to Vector"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "#telugu\n",
    "vectorizer_te = CountVectorizer()\n",
    "X = vectorizer_te.fit_transform(filtered_sentence_te)\n",
    "#print(X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "vectorizer_d = CountVectorizer()\n",
    "X_d = vectorizer_d.fit_transform(filtered_sentence_te).toarray()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Label Encoding ###"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 12, 13, 14, 15, 16, 17, 18, 19}"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#telugu\n",
    "le = LabelEncoder()\n",
    "Y = le.fit_transform(emoji_te[0:len(filtered_sentence_te)])\n",
    "la = le.inverse_transform(Y)\n",
    "set(la)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Splitting data - TELUGU"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Splitting the data into train data and test data\n",
    "X_train_te, X_test_te, y_train_te, y_test_te = train_test_split(X,Y, test_size = 0.2, random_state = 42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Splitting the data into train data and test data\n",
    "X_train_d, X_test_d, y_train_d, y_test_d = train_test_split(X_d,Y, test_size = 0.2, random_state = 42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "19\n"
     ]
    }
   ],
   "source": [
    "from tensorflow.keras.utils import to_categorical\n",
    "y_train_d = to_categorical(y_train_d)\n",
    "y_test_d = to_categorical(y_test_d)\n",
    "\n",
    "count_classes = y_test_d.shape[1]\n",
    "print(count_classes)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape train data =  (13592, 3671)\n",
      "Shape of train label =  (13592,)\n",
      "Shape of test data =  (3398, 3671)\n",
      "Shape of test label =  (3398,)\n"
     ]
    }
   ],
   "source": [
    "#Printing the size of the train data, train label, test data and test label\n",
    "print(\"Shape train data = \",np.shape(X_train_te))\n",
    "print(\"Shape of train label = \",np.shape(y_train_te))\n",
    "print(\"Shape of test data = \",np.shape(X_test_te))\n",
    "print(\"Shape of test label = \",np.shape(y_test_te))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Decision Tree###"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Time taken to train the model : 2.1174182891845703 s\n"
     ]
    }
   ],
   "source": [
    "start=time.time()\n",
    "clf_dt_te = DecisionTreeClassifier()\n",
    "clf_dt_te = clf_dt_te.fit(X_train_te, y_train_te)\n",
    "end=time.time()\n",
    "tt=end-start\n",
    "print(\"Time taken to train the model :\" ,tt,\"s\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Time taken to test the model : 0.01491999626159668 s\n"
     ]
    }
   ],
   "source": [
    "start=time.time()\n",
    "y_pred_dt_te = clf_dt_te.predict(X_test_te)\n",
    "end=time.time()\n",
    "tt=end-start\n",
    "print(\"Time taken to test the model :\" ,tt,\"s\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Decision Tree Classifier Report - Telugu"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy =  52.68\n"
     ]
    }
   ],
   "source": [
    "#Computation of the accuracy score\n",
    "acc_dt_te = accuracy_score(y_test_te, y_pred_dt_te)\n",
    "print('Accuracy = ','%.2f'%(acc_dt_te*100))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.56      0.52      0.54       196\n",
      "           1       0.47      0.58      0.52       570\n",
      "           2       0.56      0.58      0.57       752\n",
      "           3       0.63      0.57      0.60       118\n",
      "           4       0.22      0.13      0.16        47\n",
      "           5       0.55      0.48      0.51       375\n",
      "           6       0.51      0.55      0.53       430\n",
      "           7       0.75      0.67      0.71        36\n",
      "           8       0.52      0.33      0.40        52\n",
      "           9       0.51      0.39      0.44       145\n",
      "          10       0.50      0.40      0.44        75\n",
      "          11       0.73      0.55      0.63        29\n",
      "          12       0.44      0.17      0.25        23\n",
      "          13       0.53      0.47      0.50       155\n",
      "          14       0.44      0.44      0.44         9\n",
      "          15       0.49      0.51      0.50       154\n",
      "          16       0.62      0.71      0.67         7\n",
      "          17       1.00      1.00      1.00         3\n",
      "          18       0.55      0.55      0.55       222\n",
      "\n",
      "    accuracy                           0.53      3398\n",
      "   macro avg       0.56      0.50      0.52      3398\n",
      "weighted avg       0.53      0.53      0.52      3398\n",
      "\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import classification_report\n",
    "rep_dt_te = classification_report(y_test_te, y_pred_dt_te, output_dict=True)\n",
    "rep1_dt_te = classification_report(y_test_te, y_pred_dt_te)\n",
    "\n",
    "macro_precision_dt_te =  rep_dt_te['macro avg']['precision'] \n",
    "macro_recall_dt_te = rep_dt_te['macro avg']['recall']    \n",
    "macro_f1_dt_te = rep_dt_te['macro avg']['f1-score']\n",
    "accuracy_dt_te = rep_dt_te['accuracy']\n",
    "\n",
    "print(rep1_dt_te)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x1fc784258c8>"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import scikitplot as skplt\n",
    "import matplotlib.pyplot as plt\n",
    "skplt.metrics.plot_confusion_matrix(y_test_te, y_pred_dt_te)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[179, 711, 785, 106, 27, 325, 464, 32, 33, 109, 60, 22, 9, 137, 9, 159, 8, 3, 220]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAmQAAAE/CAYAAADosN8VAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8li6FKAAAgAElEQVR4nO3de3wV5bX/8c8iAbxwk0vSQNAoImAChEsRjxbrBcFCUbwBYku9Hj3aX72Lx1/rlYJWPdqCRVuxWJWIVUoKSEEQpf6qFDQCgQIKFAIIaOVwUZDL+v2xJ3EDySZIZs9O8n2/Xvu1Z56ZPWsNhM3K88w8Y+6OiIiIiESnTtQJiIiIiNR2KshEREREIqaCTERERCRiKshEREREIqaCTERERCRiKshEREREIqaCTERERCRiKshEJHQWs9LMllSwvY+ZvWNm28xss5m9bWYDzOy/zWx78NppZnvj1ovNLMfM3MzSDzjeH8zs4QPavh/se9cB7eUeo4I8DzpuZc/TzOYE57DdzP43ON+OcdvvN7Pdcee33cy2xG13Mzu5nOPWM7PHzawk+MwqM/ufQ52LiKQWFWQikgy9gAzgJDP7bvwGM7sUeBV4AcgGMoFfAD9091+6ewN3bwDcAPy9dN3dcw8zh2HAv4P3sFR4noGbg3NpBswB/njA9lfizq+BuzepRMx7gO5AD6AhcDbw4bc9ARGJhgoyEUmGYcBkYBpxBZGZGfAE8JC7/97d/9fd97n72+5+XVUFN7NjgEuBm4C2Zta9qo59gHLP80DuvgcoAE6tgpjfBSa5+3qPWe3uL1TBcUUkiVSQiUio4oqhl4LXYDOrF2xuB7QG/hRyGpcA24n1xP0V+HFVBzjEeR64bz1gKPBeFYR+D7jNzP7LzDoGRa6IVDMqyEQkbBcDu4AZwBQgHegXbGsWvG8IOYdhxIYD9wIvA0PMrG4Vx0h0nqV+HVwXth24GXjggO2Xm9mWuNdblYg7EniEWIE3H1hnZmEOy4pICFSQiUjYhgET3X2Pu+8CXueb4bzPg/esb3nsPcH7gcVVXWA3gJm1JnZd1UvBtsnAURxcLB2pROdZ6v8E14UdBfQH/mRmneK2T3T3JnGvsw8V1N33uvsYdz8DaAKMAMaZWYcqOSsRSQoVZCISGjPLBs4BrjSzT83sU2LDej8ws+bAMmAtsSHFb2MDscIr54D2E4F/Bcs/IvZd95cg/kpiBVGVDVtW4jz3E1wnNxf4GDi/qvJw96/cfQzwBVVzfZqIJIkKMhEJ04+A5cSuFcsPXqcAJcAQd3fgNuDnZnaVmTUyszpmdqaZPXuogwdDkK8BI8ysmZnVNbMhxIqRN4LdfkxsaDA/7nUJ0M/MmsUdrr6ZHRX3quj7Me2A/eod6jzLO4iZnR7kWXyo84xT74DYaWZ2SzClx9Fmlh4MVzZEd1qKVCsqyEQkTMOAp9390/gXMDbYhrv/CRgEXA2sBzYCDxMbWqyM/yI2ncVCYBOxa7P6uftGM+tJrPdszAE5FBLrnYovlrYDX8W9zqkg3vAD9ptdmfMMjC6dY4zYlBf/193fiNs+6IB5yLabWUbc9uIDYl8VvD8OfAp8RuxO0kvcfWUl//xEJAVY7BdUEREREYmKeshEREREIqaCTERERCRiKshEREREIqaCTERERCRiKshEREREIpYedQJHonnz5p6TkxN1GiIiIiKHtGDBgs/cvUV526p1QZaTk8P8+fOjTkNERETkkMzsXxVt05CliIiISMRUkImIiIhETAWZiIiISMRUkNViy5YtIz8/v+zVqFEjnnzySe6//35atWpV1j5t2jQAZs6cSbdu3ejYsSPdunVj9uzZEZ+BiIhIzVCtn2XZvXt310X9VWPv3r20atWK999/n+eff54GDRpwxx137LfPhx9+SGZmJi1btmTx4sX06dOHdevWRZSxiIhI9WJmC9y9e3nbqvVdllJ1Zs2aRZs2bTjhhBMq3KdLly5ly7m5uezcuZNdu3ZRv379ZKQoIiJSY2nIUgAoKChgyJAhZeujR4+mU6dOXH311XzxxRcH7f/aa6/RpUsXFWMiIiJVQEOWwtdff03Lli0pLi4mMzOTjRs30rx5c8yMn//852zYsIFx48aV7V9cXMyAAQOYMWMGbdq0iTBzERGR6iPRkKV6yIQ33niDrl27kpmZCUBmZiZpaWnUqVOH6667jnnz5pXtW1JSwsCBA3nhhRdUjImIiFQRFWTChAkT9huu3LBhQ9nypEmTyMvLA2DLli3069ePkSNHcsYZZyQ9TxERkZpKQ5a13Jdffknr1q1ZuXIljRs3BuBHP/oRRUVFmBk5OTk888wzZGVl8fDDDzNy5Ejatm1b9vkZM2aQkZERVfoiIiLVRqIhSxVktVjO8KlVfszVo/pV+TFFRERqAl1DJiIiIpLCVJCJiIiIREwFmYiIiEjEVJCJiIiIREwFmYiIiEjEVJCJiIiIREwFmYiIiEjEVJCJiIiIREwFmYiIiEjEVJCJiIiIREwFmYiIiEjEVJCJiIiIREwFmYiIiEjEQivIzKydmRXFvbaa2S1m1tTMZprZiuD9uLjP3GNmH5vZMjPrE1ZuIiIiIqkktILM3Ze5e7675wPdgC+BScBwYJa7twVmBeuY2anAYCAX6As8bWZpYeUnIiIikiqSNWR5LvCJu/8LuBAYH7SPBy4Kli8ECtx9l7uvAj4GeiQpPxEREZHIJKsgGwxMCJYz3X0DQPCeEbS3AtbGfaYkaBMRERGp0UIvyMysHjAAePVQu5bT5uUc73ozm29m8zdv3lwVKYqIiIhEKhk9ZBcAH7j7xmB9o5llAQTvm4L2EqB13OeygfUHHszdn3X37u7evUWLFiGmLSIiIpIcySjIhvDNcCVAITAsWB4GTI5rH2xm9c3sRKAtMC8J+YmIiIhEKj3Mg5vZMUBv4D/jmkcBE83sGmANcBmAuxeb2URgCbAHuMnd94aZn4iIiEgqCLUgc/cvgWYHtH1O7K7L8vYfAYwIMycRERGRVKOZ+kVEREQipoJMREREJGIqyEREREQipoJMREREJGIqyEREREQipoJMREREJGIqyEREREQipoJMREREJGIqyEREREQipoJMREREJGIqyEREREQipoJMREREJGIqyEREREQipoJMREREJGIqyEREREQipoJMREREJGIqyEREREQipoJMREREJGIqyEREREQipoJMREREJGIqyEREREQipoJMREREJGIqyEREREQipoJMREREJGIqyEREREQiFmpBZmZNzOxPZvZPM1tqZqebWVMzm2lmK4L34+L2v8fMPjazZWbWJ8zcRERERFJF2D1kTwHT3b090BlYCgwHZrl7W2BWsI6ZnQoMBnKBvsDTZpYWcn4iIiIikQutIDOzRkAv4DkAd//a3bcAFwLjg93GAxcFyxcCBe6+y91XAR8DPcLKT0RERCRVhNlDdhKwGXjezD40s9+b2bFAprtvAAjeM4L9WwFr4z5fErSJiIiI1GhhFmTpQFfgt+7eBdhBMDxZASunzQ/ayex6M5tvZvM3b95cNZmKiIiIRCjMgqwEKHH394P1PxEr0DaaWRZA8L4pbv/WcZ/PBtYfeFB3f9bdu7t79xYtWoSWvIiIiEiyhFaQufunwFozaxc0nQssAQqBYUHbMGBysFwIDDaz+mZ2ItAWmBdWfiIiIiKpIj3k4/8UeMnM6gErgauIFYETzewaYA1wGYC7F5vZRGJF2x7gJnffG3J+KSknJ4eGDRuSlpZGeno68+fPZ9CgQSxbtgyALVu20KRJE4qKigAYOXIkzz33HGlpafz617+mTx/NGCIiIlKdhFqQuXsR0L2cTedWsP8IYESYOVUXb731Fs2bNy9bf+WVV8qWb7/9dho3bgzAkiVLKCgooLi4mPXr13PeeeexfPly0tI0Y4iIiEh1oZn6qxl3Z+LEiQwZMgSAyZMnM3jwYOrXr8+JJ57IySefzLx5GukVERGpTlSQpSAz4/zzz6dbt248++yz+22bO3cumZmZtG3bFoB169bRuvU390JkZ2ezbt26pOYrIiIiRybsa8jkW3j33Xdp2bIlmzZtonfv3rRv355evXoBMGHChLLeMYj1mB3IrLwZRERERCRVqYcsBbVs2RKAjIwMBg4cWDYEuWfPHl5//XUGDRpUtm92djZr134zn25JSUnZ50VERKR6UEGWYnbs2MG2bdvKlmfMmEFeXh4Ab775Ju3btyc7O7ts/wEDBlBQUMCuXbtYtWoVK1asoEcPPXFKRESkOtGQZYrZuHEjAwcOBGI9YldccQV9+/YFoKCgYL/hSoDc3Fwuv/xyTj31VNLT0xkzZozusBQREalmrLxrkKqL7t27+/z586NOo0rlDJ9a5cdcPapf5LFERERqOzNb4O7lTQemIUsRERGRqKkgExEREYmYCjIRERGRiKkgExEREYmYCjIRERGRiKkgExEREYmYCjIRERGRiKkgExEREYmYCjIRERGRiKkgExEREYmYCjIRERGRiKkgExEREYmYCjIRERGRiKkgExEREYmYCjIRERGRiB1WQWZmbczs/5rZ4rASEhEREaltDlmQmVmWmd1iZvOAYiANGBJ6ZiIiIiK1RIUFmZldZ2azgbeB5sC1wAZ3f8DdFyUrQREREZGaLj3BtjHA34Er3H0+gJl5UrISERERqUUSDVm2BAqAJ8xsmZk9BNQ9nIOb2WozW2RmRWZWWtQ1NbOZZrYieD8ubv97zOzjIF6fb3NCIiIiItVNhQWZu3/m7r91917AucD/ApvMbKmZ/fIwYpzt7vnu3j1YHw7Mcve2wKxgHTM7FRgM5AJ9gafNLO3wT0lERESkekl0DVnP0mV3L3H3x9y9G3ARsOsIYl4IjA+WxwfHK20vcPdd7r4K+BjocQRxRERERKqFREOWT5fX6O7L3P2BSh7fgRlmtsDMrg/aMt19Q3CsDUBG0N4KWBv32ZKgTURERKRGS3RRf1U4w93Xm1kGMNPM/plgXyun7aCbCILC7nqA448/vmqyFBEREYlQooLsJDMrrGijuw841MHdfX3wvsnMJhEbgtxoZlnuvsHMsoBNwe4lQOu4j2cD68s55rPAswDdu3fXXZ8iIiJS7SUqyDYDj3/bA5vZsUAdd98WLJ8PPAgUAsOAUcH75OAjhcDLZvYEsTs82wLzvm18ERERkeoiUUG2zd3fPoJjZwKTzKw0zsvuPt3M/gFMNLNrgDXAZQDuXmxmE4ElwB7gJnffewTxRURERKqFRAXZ6oo2mFldd9+d6MDuvhLoXE7758Sm0SjvMyOAEYmOKyIiIlLTJJqH7OL4dYs5x8x+T+x6LxERERGpApV5uPhpZvYU8C9i13nNBdqHnZiIiIhIbZFoYtgRZrYC+CWwCOgCbHb38e7+RbISFBEREanpEl1Ddj2wDPgtMMXdd+rh4iIiIiJVL9GQ5XeIXWA/APjYzP4IHG1mYU8mKyIiIlKrVFhcBVNOvAG8YWZHAf2BY4B1ZjbL3a9IUo4iIiIiNVqlervcfSfwJ+BPZtYIGBhqViIiIiK1SMKCzMzaEbuWrPSuyqXAs+4+PuzERERERGqLRHdZng7MAbYRe3bk74AdwBwz65mU7ERERERqgUQ9ZL8Ahrj7nLi2P5vZbOA+4IIwExMRERGpLRLdZdnmgGIMgOD5lieFlpGIiIhILZOoINuWYNuOqk5ERCpn7969dOnShf79+wPw85//nE6dOpGfn8/555/P+vXry/YdOXIkJ598Mu3ateOvf/1rVCmLiMghJBqybG1mvy6n3YBWIeUjIofw1FNP0aFDB7Zu3QrAnXfeyUMPPQTAr3/9ax588EHGjh3LkiVLKCgooLi4mPXr13PeeeexfPly0tLSokxfRETKkaiH7E5gQTmv+cBd4acmIgcqKSlh6tSpXHvttWVtjRo1KlvesWMHZgbA5MmTGTx4MPXr1+fEE0/k5JNPZt68eUnPWUREDi3RxLCa2kIkxdxyyy08+uijbNu2/xUF9957Ly+88AKNGzfmrbfeAmDdunX07PnNDdHZ2dmsW7cuqfmKiEjlJOohE5EUMmXKFDIyMujWrdtB20aMGMHatWsZOnQoo0ePBsD94EfPlvaeiYhIalFBJlJNvPvuuxQWFpKTk8PgwYOZPXs2V1555X77XHHFFbz22mtArEds7dq1ZdtKSkpo2bJlUnMWEZHKSTQx7M3JTEREEhs5ciQlJSWsXr2agoICzjnnHF588UVWrFhRtk9hYSHt28cerDFgwAAKCgrYtWsXq1atYsWKFfTo0SOq9EVEJIFEd1leDYxOViIi8u0MHz6cZcuWUadOHU444QTGjh0LQG5uLpdffjmnnnoq6enpjBkzRndYioikqEo9XFxEopczfOr+DXk3xtraXg1tY02LgDN+UwQUBTvlw2W/Zi9w49v74O39j7F6VL+QsxYRkcpIVJB1MrOt5bQb4O7eqJxtIiIiInKYEhVki9y9S9IyEREREamldJeliIiISMQSFWSvJi0LERERkVosUUG22czaAljM82a21cwWmlnXJOUnIiIiUuMlKsh+BqwOlocAnYATgduApyobwMzSzOxDM5sSrDc1s5lmtiJ4Py5u33vM7GMzW2ZmfQ73ZERERESqo0QF2R533x0s9wdecPfP3f1N4NjDiPEzYGnc+nBglru3BWYF65jZqcBgIBfoCzxtZpo0SURERGq8RAXZPjPLMrOjgHOBN+O2HV2Zg5tZNtAP+H1c84VA6YPLxwMXxbUXuPsud18FfAxoWnERERGp8RIVZL8A5hMbtix092IAMzsLWFnJ4z8J3AXsi2vLdPcNAMF7RtDeClgbt19J0CYiIiJSo1U4D5m7TzGzE4CG7v5F3Kb5wKBDHdjM+gOb3H2BmX2/ErlYeWmUc9zrgesBjj/++EocVkRERCS1JXq4+F3uvsfdvzCzy0rb3X0H8N+VOPYZwAAzWw0UAOeY2YvARjPLCmJkAZuC/UuA1nGfzwbWH3hQd3/W3bu7e/cWLVpUIg2R8OzcuZMePXrQuXNncnNzue+++wAYNGgQ+fn55Ofnk5OTQ35+PgCrV6/m6KOPLtt2ww03RJm+iIikiEQz9Q8GHg2W72H/ecn6coiizN3vCT5H0EN2h7tfaWa/AoYBo4L3ycFHCoGXzewJoCWxp/PNO5yTEUm2+vXrM3v2bBo0aMDu3bs588wzueCCC3jllVfK9rn99ttp3Lhx2XqbNm0oKioq73AiIlJLJSrIrILl8tYPxyhgopldA6wBLgNw92IzmwgsAfYAN7n73iOIIxI6M6NBgwYA7N69m927d2P2zT8Pd2fixInMnj07qhRFRKQaSHRRv1ewXN56Qu4+x937B8ufu/u57t42eP933H4j3L2Nu7dz9zcOJ4ZIVPbu3Ut+fj4ZGRn07t2b0047rWzb3LlzyczMpG3btmVtq1atokuXLpx11lnMnTs3ipRFRCTFJOoh62xmW4n1hh0dLBOsHxV6ZiLVRFpaGkVFRWzZsoWBAweyePFi8vLyAJgwYQJDhgwp2zcrK4s1a9bQrFkzFixYwEUXXURxcTGNGjWKKn0REUkBiXrIjnL3Ru7e0N3Tg+XS9bpJy1CkmmjSpAnf//73mT59OgB79uzh9ddfZ9Cgb25Krl+/Ps2aNQOgW7dutGnThuXLl0eSr4iIpI5EBdn7SctCpJravHkzW7ZsAeCrr77izTffpH379gBly9nZ2fvtv3dv7NLIlStXsmLFCk466aTkJy4iIimlshf1i0g5NmzYwLBhw9i7dy/79u3j8ssvp3///gAUFBTsN1wJ8M477/CLX/yC9PR00tLSGDt2LE2bNo0idRERSSGJCrIWZnZbRRvd/YkQ8hGpVga8vBb6PFy2Pu5LGDd8amzlO5cxZzWMKl0H4Cj44aNlaz99F376bvx2WD2qX4gZi4hIKkpUkKUBDVBPmYiIiEioEhVkG9z9waRlIiIiIlJLJbqoXz1jIiIiIkmQqCA7N2lZiIiIiNRiFRZk8TPoi4iIiEh4EvWQiYiIiEgSVFiQmVnPZCYiIiIiUlsl6iF7unTBzP6ehFxEREREaqXK3mWph4mLiIiIhCTRPGR1zOw4YkVb6XJZkaaL/kVERESqRqKCrDGwgG+KsA/itjmgJyKLiIiIVIEKCzJ3z0liHiIiIiK1VqIeMsysHjAUyCXWK7YEeNnddyUhNxEREZFaIdG0F6cSK8C+D6wBSoLlYjPLTUZyIiIiIrVBoh6y3wA3uvvM+EYzOw8YDZwdZmIiIiIitUWiaS9aHViMAbj7m8B3wktJREREpHZJVJDVMbP6Bzaa2VEc4tozEREREam8RAXZC8BrZpZT2hAsTwT+GGZSIiIiIrVJomkvHjazm4F3zOyYoHkH8Ji7/yYp2YmIiIjUAgmHHt19NDDazBoG69uSkpWIiIhILVKpa8FUiImIiIiEJ9E1ZEfEzI4ys3lm9pGZFZvZA0F7UzObaWYrgvfj4j5zj5l9bGbLzKxPWLmJiIiIpJLQCjJgF3COu3cG8oG+ZtYTGA7Mcve2wKxgvXQi2sHEngrQF3jazNJCzE9EREQkJRzq0UkZwE3s/+ikp91946EO7O4ObA9W6wYvBy4kNuM/wHhgDnB30F4QPJZplZl9DPQA/n5YZyQiIiJSzSR6dNIZwD+C1ReAF4Pl94Nth2RmaWZWBGwCZrr7+0Cmu28ACN4zgt1bAWvjPl4StImIiIjUaIl6yB4HLnL3D+PaJpvZJOAZ4LRDHdzd9wL5ZtYEmGRmeQl2t/IOcdBOZtcD1wMcf/zxh0pBREREJOUluoas0QHFGADuXgQ0PJwg7r6F2NBkX2CjmWUBBO+bgt1KgNZxH8sG1pdzrGfdvbu7d2/RosXhpCEiIiKSkhIVZBZ/B2RcY9NDfK50vxZBzxhmdjRwHvBPoBAYFuw2DJgcLBcCg82svpmdCLQF5lX2RERERESqq0RDlv8DzDCzO4APgrZuwCPBtkPJAsYHd0rWASa6+xQz+zsw0cyuAdYAlwG4e7GZTSR248Ae4KZgyFNERESkRkv06KRnzWw98BCxuywBioGH3f0vhzqwuy8EupTT/jlwbgWfGQGMqETeIiIiIjXGoR6dNAWYkqRcRERERGqlCgsyM/tFgs+5uz8UQj4iIiIitU6iHrId5bQdC1wDNCM2lCkiIiIiRyjRNWSPly6bWUPgZ8BVQAGxOcpEREREpAoc6tFJTYHbgKHEHnPU1d2/SEZiIiIiIrVFomvIfgVcDDwLdHT37RXtKyIiIiLfXqIJXm8HWgL/F1hvZluD1zYz25qc9ERERERqvkTXkB1yNn4REREROXIqukREREQipoJMREREJGIqyEREREQipoJMREREJGIqyEREREQipoJMREREJGIqyERERKRaWbt2LWeffTYdOnQgNzeXp556CoA777yT9u3b06lTJwYOHMiWLVsAmDlzJt26daNjx45069aN2bNnR5l+uVSQiYiISLWSnp7O448/ztKlS3nvvfcYM2YMS5YsoXfv3ixevJiFCxdyyimnMHLkSACaN2/OX/7yFxYtWsT48eP50Y9+FPEZHEwFmYiIiFQrWVlZdO3aFYCGDRvSoUMH1q1bx/nnn096emzO+549e1JSUgJAly5daNmyJQC5ubns3LmTXbt2RZN8BVSQiYiISLW1evVqPvzwQ0477bT92seNG8cFF1xw0P6vvfYaXbp0oX79+slKsVIqfHSSiIiISCrbvn07l1xyCU8++SSNGjUqax8xYgTp6ekMHTp0v/2Li4u5++67mTFjRrJTPSQVZCIiIlLt7N69m0suuYShQ4dy8cUXl7WPHz+eKVOmMGvWLMysrL2kpISBAwfywgsv0KZNmyhSTkgFmYiIiFQr7s4111xDhw4duO2228rap0+fziOPPMLbb7/NMcccU9a+ZcsW+vXrx8iRIznjjDOiSPmQVJCJiIhItZEzfCo7S4rZ+NIfqdsih6cnFAJwXK8f8+83n8X37qZVXux6svot29Gsz81s+X8FbF26jCH/dSf8150AZF7+EGnHNik77upR/ZJ/MnFUkImIiEi1clR2LifcPeWg9lZtvlvu/k3+YzBN/mNw2GkdEd1lWUlXX301GRkZ5OXllbV99NFHnH766XTs2JEf/vCHbN26tWzbwoULOf3008nNzaVjx47s3LkzirRFRESkGlBBVkk/+clPmD59+n5t1157LaNGjWLRokUMHDiQX/3qVwDs2bOHK6+8krFjx1JcXMycOXOoW7duFGmLiIhINRBaQWZmrc3sLTNbambFZvazoL2pmc00sxXB+3Fxn7nHzD42s2Vm1ies3L6NXr160bRp0/3ali1bRq9evQDo3bs3r732GgAzZsygU6dOdO7cGYBmzZqRlpaW3IRFRESk2gizh2wPcLu7dwB6AjeZ2anAcGCWu7cFZgXrBNsGA7lAX+BpM0vpKiYvL4/CwtjFhK+++ipr164FYPny5ZgZffr0oWvXrjz66KNRpikiIiIpLrSCzN03uPsHwfI2YCnQCrgQGB/sNh64KFi+EChw913uvgr4GOgRVn5VYdy4cYwZM4Zu3bqxbds26tWrB8SGLP/2t7/x0ksv8be//Y1JkyYxa9asiLMVERGRVJWUa8jMLAfoArwPZLr7BogVbUBGsFsrYG3cx0qCtpTVvn17ZsyYwYIFCxgyZEjZRHPZ2dmcddZZNG/enGOOOYYf/OAHfPDBBxFnKyIiIqkq9ILMzBoArwG3uPvWRLuW0+blHO96M5tvZvM3b95cVWl+K5s2bQJg3759PPzww9xwww0A9OnTh4ULF/Lll1+yZ88e3n77bU499dQoUxUREZEUFuo8ZGZWl1gx9pK7vx40bzSzLHffYGZZwKagvQRoHffxbGD9gcd092eBZwG6d+9+UMEWhpzhU9lc+Ci71ixi71dbSW/YnMZnDsV3f8W2D6YCcMwp/8HbWRk8ODy2vr3VOTQ5vgMYHH1Sd26aCzfNnVp2zKgnoBMREZHUEVpBZrEHSD0HLHX3J+I2FQLDgFHB++S49pfN7AmgJdAWmBdWfoerxYC7ym1v1P3Cctsb5CcUBUkAABfqSURBVJ5Ng9yzw0xJREREaogwe8jOAH4ELDKzoqDtv4kVYhPN7BpgDXAZgLsXm9lEYAmxOzRvcve9IeYnIiIikhJCK8jc/W+Uf10YwLkVfGYEMCKsnERERERSkWbqFxEREYmYCjIRERGRiKkgExEREYmYCjIREUlZV199NRkZGeTl5ZW13X///bRq1Yr8/Hzy8/OZNm0aAKtXr+boo48uay+dGzIVHc55lVqzZg0NGjTgscceS3a6kgQqyEREJGX95Cc/Yfr06Qe133rrrRQVFVFUVMQPfvCDsvY2bdqUtY8dOzaZqR6Wwz2v0m0XXHBBslKUJAt1YlgREZEj0atXL1avXh11GlXucM/rz3/+MyeddBLHHntseElJpNRDJiK1SnlDRaUee+wxzIzPPvsMqF5DYLXN6NGj6dSpE1dffTVffPFFWfuqVavo0qULZ511FnPnzo0ww2+nvPPasWMHjzzyCPfdd1/E2UmYVJCJSK1S0VDR2rVrmTlzJscff/x+7dVlCKw2ufHGG/nkk08oKioiKyuL22+/HYCsrCzWrFnDhx9+yBNPPMEVV1zB1q2JHqGcWio6r/vuu49bb72VBg0aRJyhhEkFmYjUKr169aJp06YHtd966608+uijxJ76JqksMzOTtLQ06tSpw3XXXce8ebGn7NWvX59mzZoB0K1bN9q0acPy5cujTPWwVHRe77//PnfddRc5OTk8+eST/PKXv2T06NERZytVTdeQiUitV1hYSKtWrejcufNB20qHwBo1asTDDz/M9773vQgylHgbNmwgKysLgEmTJpUNP2/evJmmTZuSlpbGypUrWbFiBSeddFKUqR6Wis4rfuj1/vvvp0GDBtx8882R5CjhUUEmIrXal19+yYgRI5gxY8ZB20qHwJo1a8aCBQu46KKLKC4uplGjRhFkWrvkDJ8KwObCR9m1ZhF7v9pKesPmND5zKLvWLuLrjSvBjPTGGTTtczM5w6eyY9m7/O/cl6BOHaxOGo3PvIauj/59v+OuHtUvitM5yJAhQ5gzZw6fffYZ2dnZPPDAA8yZM4eioiLMjJycHJ555pmo05QkUkEmIrXaJ598wqpVq8p6x0pKSujatSvz5s3jO9/5DvXr1wf2HwLr3r17lCnXKi0G3HVQW8PO55e777HtzuDYdmeEndIRKS00OeFK6g+7klZB+0MrgFaDYy9gIXD6Ux+Uc4Tvwk4YXXqcQKoUmvLtqSATkVqtY8eObNq0qWw9JyeH+fPn07x582o/BCYi1Ycu6heRWmXIkCGcfvrpLFu2jOzsbJ577rkK933nnXfo1KkTnTt35tJLL2Xs2LHl3hAgInKk1EMmIrVGzvCp5Q4VPRQ//DN4DN0fez9YOQp++GjZpp++Cz99V0NFIlL11EMmSVPehJyvvvoqubm51KlTh/nz55e17969m2HDhtGxY0c6dOjAyJEjo0hZREQkKVSQSdKUNyFnXl4er7/+Or169dqv/dVXX2XXrl0sWrSIBQsW8Mwzz9TIx6eIiIiAhiwlicp7dluHDh3K3dfM2LFjB3v27OGrr76iXr16mmpARERqLPWQSUq69NJLOfbYY8nKyuL444/njjvu0MXUIiJSY6mHTFLSvHnzSEtLY/369XzxxRd873vf47zzztOUAyIiUiOph0xS0ssvv0zfvn2pW7cuGRkZnHHGGftd9C8iIlKTqCCTlHT88ccze/Zs3J0dO3bw3nvv0b59+6jTEhERCYUKMkma8ibknDRpEtnZ2fz973+nX79+9OnTB4CbbrqJ7du3k5eXx3e/+12uuuoqOnXqFPEZiIiIhEPXkElSVDQhJ0D6lc+QFbQtI+5Zb21+Am1ii2M+hzF6dpuIiNRQ6iETERERiZgKMhEREZGIhVaQmdk4M9tkZovj2pqa2UwzWxG8Hxe37R4z+9jMlplZn7DyEhEREUk1YfaQ/QHoe0DbcGCWu7cFZgXrmNmpwGAgN/jM02aWFmJuIiIiIikjtILM3d8B/n1A84XA+GB5PHBRXHuBu+9y91XAx0CPsHITERERSSXJvoYs0903AATvGUF7K2Bt3H4lQdtBzOx6M5tvZvM3b94carIiIiIiyZAqF/VbOW1e3o7u/qy7d3f37i1atAg5LRGRb2fZsmXk5+eXvRo1asSTTz7J/fffT6tWrcrap02bFnWqIpICkj0P2UYzy3L3DWaWBWwK2kuA1nH7ZQPrk5ybiEiVadeuHUVFRQDs3buXVq1aMXDgQJ5//nluvfVW7rjjjogzFJFUkuweskJgWLA8DJgc1z7YzOqb2YlAW2BeknMTEQnFrFmzaNOmDSeccELUqYhIigpz2osJwN+BdmZWYmbXAKOA3ma2AugdrOPuxcBEYAkwHbjJ3feGlZuISDIVFBQwZMiQsvXRo0fTqVMnrr76ar744osIMxORVBHmXZZD3D3L3eu6e7a7P+fun7v7ue7eNnj/d9z+I9y9jbu3c/c3wspLRCSZvv76awoLC7nssssAuPHGG/nkk08oKioiKyuL22+/PeIMRSQVpMpF/SIiNdIbb7xB165dyczMBCAzM5O0tDTq1KnDddddx7x5ujpDRFSQiYiEasKECfsNV27YsKFsedKkSeTl5UWRloikmGTfZSkiUmt8+eWXzJw5k2eeeaas7a677qKoqAgzIycnZ79tIlJ7qSATkcjl5OTQsGFD0tLSSE9PZ/78+dx555385S9/oV69erRp04bnn3+eJk2aRJ1qpeUMnwpAw+teoPPIv32zodXg2AtYCJz+1AeVPubqUf2qMkURSSEashSRlPDWW29RVFTE/PnzAejduzeLFy9m4cKFnHLKKYwcOTLiDEVEwqOCTGqsvXv30qVLF/r37w/AoEGDymZHz8nJIT8/P+IMJZHzzz+f9PRYJ37Pnj0pKSmJOCMRkfBoyFJqrKeeeooOHTqwdetWAF555ZWybbfffjuNGzeOKjU5gJlx/vnnY2b853/+J9dff/1+28eNG8egQYMiyk5EJHzqIZMaqaSkhKlTp3LttdcetM3dmThx4n53vkm03n33XT744APeeOMNxowZwzvvvFO2bcSIEaSnpzN06NAIMxQRCZcKMqmRbrnlFh599FHq1Dn4R3zu3LlkZmbStm3bCDKT8rRs2RKAjIwMBg4cWDY31/jx45kyZQovvfQSZhZliiIioVJBJjXOlClTyMjIoFu3buVuP3BeKDnYzp076dGjB507dyY3N5f77ruvbNtvfvMb2rVrR25uLnfdddcRx9qxYwfbtm0rW54xYwZ5eXlMnz6dRx55hMLCQo455pgjjiMiksp0DZnUOO+++y6FhYVMmzaNnTt3snXrVq688kpefPFF9uzZw+uvv86CBQuiTjOl1a9fn9mzZ9OgQQN2797NmWeeyQUXXMBXX33F5MmTWbhwIfXr12fTpk1HHGvjxo0MHDgQgD179nDFFVfQt29fTj75ZHbt2kXv3r2B2IX9Y8eOPeJ4IiKpSAWZ1DgjR44smyJhzpw5PPbYY7z44osAvPnmm7Rv357s7OwoU0x5ZkaDBg0A2L17N7t378bM+O1vf8vw4cOpX78+EBtiPFLnPLsULvhl2frvtsHvhk+FS58iDdgStE/nm7m9DqW2zde1du1afvzjH/Ppp59Sp04drr/+en72s5/x0UcfccMNN7B9+3ZycnJ46aWXaNSoUdTpikg5NGQptUpBQYGGKytp79695Ofnk5GRQe/evTnttNNYvnw5c+fO5bTTTuOss87iH//4R9RpCpCens7jjz/O0qVLee+99xgzZgxLlizh2muvZdSoUSxatIiBAwfyq1/9KupURaQC6iGTGuegXpS8G79p+85lzFkNoyrZ01KqtvW4AKSlpVFUVMSWLVsYOHAgixcvZs+ePXzxxRe89957/OMf/+Dyyy9n5cqVuuA+YllZWWRlZQHQsGFDOnTowLp161i2bBm9evUCYhPt9unTh4ceeuiIYlXUG/fzn/+cyZMnU6dOHTIyMvjDH/5QdrOGiByaeshEJKEmTZrw/e9/n+nTp5Odnc3FF1+MmdGjRw/q1KnDZ599FnWKEmf16tV8+OGHnHbaaeTl5VFYWAjAq6++ytq1a4/4+BX1xt15550sXLiQoqIi+vfvz4MPPnjEsURqExVkInKQzZs3s2VL7Oqtr776quzau4suuojZs2cDsHz5cr7++muaN28eZaoSZ/v27VxyySU8+eSTNGrUiHHjxjFmzBi6devGtm3bqFev3hHHyMrKomvXrsD+vXHx16bt2LFDvaYih0lDliJykM53v8JnU/8HfB/4Po5p/z1u/pvhe7/D59NeYdRvT8DS6tLk7Bs58Z5plTpmbRz2Tabdu3dzySWXMHToUC6++GIA2rdvz4wZM4BYAT116uEN1R9KfG8cwL333ssLL7xA48aNeeutt6o0lkhNp4JMRA5SL+NEWl7164PaLa0uzX94RwQZSSLuzjXXXEOHDh247bbbyto3bdpERkYG+/bt4+GHH+aGG26ospgH9sZB7KkKI0aMYOTIkYwePZoHHnigyuKJ1HQashQRqebeffdd/vjHPzJ79mzy8/PJz89n2rRpTJgwgVNOOYX27dvTsmVLrrrqqiqJV15vXLwrrriC1157rUpiJdNTTz1FXl4eubm5PPnkk1GnI7WMeshERKqx0juIT7h7Clvj2v/rHQdOhov/B4ACoKCSw8tQ8RBzRb1xK1asKHscWWFhIe3btz+s84ja4sWL+d3vfse8efOoV68effv2pV+/ftX+EWvLli1j0KBBZesrV67kwQcf5JZbbokwKymPCjIREam00t64jh07kp+fD8Avf/lLnnvuOZYtW0adOnU44YQTqt1TFZYuXUrPnj3LHtN11llnMWnSpCp5PFiU2rVrR1FRERCbW7BVq1ZlT8aoajt37qRXr17s2rWLPXv2cOmll2rY+jCoIBMRkUpJ2BvX9moIOpMWAWf8pggoqtRxU+GGj7y8PO69914+//xzjj76aKZNm0b37t2jTqtKzZo1izZt2nDCCSeEcvyKHrnWs2fPUOLVNCrIRESk1uvQoQN33303vXv3pkGDBnTu3Jn09Jr1X2TYTyqp6JFrUjm6qF9ERAS45ppr+OCDD3jnnXdo2rRptb9+LN7XX39NYWEhl112WahxynvkmlSOCjIRERFi04QArFmzhtdff71GPff2jTfeoGvXrmRmZoYap/SRayUlJcybN4/FixeHGq8mSbn+WDPrCzwFpAG/d/dREackIiK1wCWXXMLnn39O3bp1GTNmDMcdd1zUKVWZCRMmJLXAjH/kWl5eXtLiVmcpVZCZWRowBugNlAD/MLNCd18SbWYiIlJTld6swBnDy9qumbkTZh7Zkw2ivlmh9Lz27d7JusnT+H9ZF/PA8CN/WkNF57V582bq1q1LkyZNyh65dvfddx9xvNoipQoyoAfwsbuvBDCzAuBCQAWZiIjIt1Cn7lG0/tmE0ONs2LCBYcOGsXfvXvbt28fll19O//79Q49bU6RaQdYKWBu3XgLoikAREZEUVtbL2OfhsrZxX8K4I+yRi7qXMZnM3aPOoYyZXQb0cfdrg/UfAT3c/adx+1wPXB+stgOWJT3RijUHPlMsxVIsxVIsxVKsWherMk5w9xblbUi1HrISoHXcejawPn4Hd38WeDaZSVWWmc1396TMJKhYiqVYiqVYiqVYqRPrSKXatBf/ANqa2YlmVg8YDBRGnJOIiIhIqFKqh8zd95jZzcBfiU17Mc7diyNOS0RERCRUKVWQAbj7NGBa1Hl8S8kcSlUsxVIsxVIsxVKs1Il1RFLqon4RERGR2ijVriETERERqXVUkFURM+trZsvM7GMzG37oT3zrOOPMbJOZhf6AMDNrbWZvmdlSMys2s5+FGOsoM5tnZh8FsR4IK1ZczDQz+9DMpoQcZ7WZLTKzIjObH3KsJmb2JzP7Z/D3dnpIcdoF51P62mpmt4QRK4h3a/BzsdjMJpjZUSHG+lkQp7iqz6m8f79m1tTMZprZiuC9Sp7XU0Gsy4Lz2mdmVXbnWQWxfhX8HC40s0lm1iTEWA8FcYrMbIaZtQwrVty2O8zMzax5WLHM7H4zWxf37+wHYcUK2n8a/D9WbGaPhhXLzF6JO6fVZlYUYqx8M3uv9PvXzHpURaxQuLteR/gidgPCJ8BJQD3gI+DUkGL1AroCi5NwXllA12C5IbA8xPMyoEGwXBd4H+gZ8vndBrwMTAk5zmqgedh/X0Gs8cC1wXI9oEkSYqYBnxKbXyeM47cCVgFHB+sTgZ+EFCsPWAwcQ+wa2zeBtlV4/IP+/QKPAsOD5eHAIyHG6kBs/sY5QPeQz+t8ID1YfiTk82oUt/x/gLFhxQraWxO7+exfVfVvu4Lzuh+4o6r+ng4R6+zg571+sJ4R5p9h3PbHgV+EeF4zgAuC5R8Ac6r6z7OqXuohqxplj3xy96+B0kc+VTl3fwf4dxjHLifWBnf/IFjeBiwl9p9jGLHc3bcHq3WDV2gXOJpZNtAP+H1YMZLNzBoR+0J6DsDdv3b3LUkIfS7wibv/K8QY6cDRZpZOrFhaf4j9v60OwHvu/qW77wHeBgZW1cEr+Pd7IbFCmuD9orBiuftSd6/yybQriDUj+DMEeI/YvJJhxdoat3osVfTdkeD79n+Au6oqziFiVbkKYt0IjHL3XcE+m0KMBYCZGXA5UCXPdaoglgONguXGhPfdccRUkFWN8h75FErhEhUzywG6EOu5CitGWtB1vQmY6e6hxQKeJPaFui/EGKUcmGFmCyz2pImwnARsBp4PhmJ/b2bHhhiv1GCq6Au1PO6+DngMWANsAP7X3WeEFG4x0MvMmpnZMcR+o259iM8cqUx33wCxX4KAjJDjReFq4I0wA5jZCDNbCwwFfhFinAHAOnf/KKwYB7g5GI4dV1XD2RU4Bfiemb1vZm+b2XdDjFXqe8BGd18RYoxbgF8FPxuPAfeEGOuIqCCrGlZOW425fdXMGgCvAbcc8JtolXL3ve6eT+w36R5mlhdGHDPrD2xy9wVhHL8cZ7h7V+AC4CYz6xVSnHRi3fW/dfcuwA5iQ2ChsdgEzgOAV0OMcRyxXqQTgZbAsWZ2ZRix3H0pseG1mcB0Ypcf7En4IUnIzO4l9mf4Uphx3P1ed28dxLk5jBhBkX4vIRZ8B/gt0AbIJ/bLyOMhxkoHjgN6AncCE4MerDANIcRf5gI3ArcGPxu3EowgpCIVZFXjkI98qq7MrC6xYuwld389GTGDYbY5QN+QQpwBDDCz1cSGl88xsxdDioW7rw/eNwGTiA1xh6EEKInrWfwTsQItTBcAH7j7xhBjnAescvfN7r4beB34j7CCuftz7t7V3XsRG/4I87d3gI1mlgUQvFfJUFEqMLNhQH9gqAcX8STBy8AlIR27DbFfDD4Kvj+ygQ/M7DthBHP3jcEvqvuA3xHedwfEvj9eDy4fmUds9KBKblgoT3D5wcXAK2HFCAwj9p0BsV8cU/aifhVkVaNGPvIp+O3oOWCpuz8RcqwWpXdhmdnRxP4T/mcYsdz9HnfPdvccYn9Xs909lB4XMzvWzBqWLhO70DmUO2Td/VNgrZm1C5rOBZaEEStOMn7DXQP0NLNjgp/Jc4ldzxgKM8sI3o8n9h9G2OdXSOw/DYL3ySHHSwoz6wvcDQxw9y9DjtU2bnUA4X13LHL3DHfPCb4/Sojd+PRpGPFKC/XAQEL67gj8GTgniHsKsZuCwnwo93nAP929JMQYEOscOStYPofwf8H69qK+q6CmvIhda7Kc2N2W94YYZwKxruvdxL4Mrgkx1pnEhl4XAkXB6wchxeoEfBjEWkwV3XVTibjfJ8S7LIld1/VR8CoO82cjiJcPzA/+HP8MHBdirGOAz4HGSfh7eoDYf7KLgT8S3AkWUqy5xArZj4Bzq/jYB/37BZoBs4j9RzELaBpirIHB8i5gI/DXEGN9TOza2tLvjqq687G8WK8FPxsLgb8ArcKKdcD21VTdXZblndcfgUXBeRUCWSHGqge8GPw5fgCcE+afIfAH4IaqiHGI8zoTWBD8e34f6FaVMavypZn6RURERCKmIUsRERGRiKkgExEREYmYCjIRERGRiKkgExEREYmYCjIRERGRiKkgExEREYmYCjIRERGRiKkgExEREYnY/wdWDKi+liGxQgAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#ax = plt.subplot(1,2,1)\n",
    "fig = plt.figure(figsize = (10, 5)) \n",
    "x = np.arange(19)\n",
    "height = []\n",
    "for i in range(0,19):\n",
    "    height.append(len(y_pred_dt_te[y_pred_dt_te==i]))\n",
    "print(height)\n",
    "plt.bar(x, height = height)\n",
    "plt.xticks(x, x)\n",
    "plt.title(\" PREDICTED LABELS\")\n",
    "plt.ylabel(\"NO OF TEST DATA\")\n",
    "for index, value in enumerate(x):\n",
    "    \n",
    "    plt.text(value-0.4,len(y_pred_dt_te[y_pred_dt_te==value]), str(len(y_pred_dt_te[y_pred_dt_te==value])))\n",
    "fig1 = plt.figure(figsize = (10, 5)) \n",
    "height=[]\n",
    "for i in range(0,19):\n",
    "    height.append(len(y_test_te[y_test_te==i]))\n",
    "plt.bar(x, height = height)\n",
    "plt.xticks(x, x)\n",
    "plt.title(\" ACTUAL LABELS\")\n",
    "plt.ylabel(\"NO OF TEST DATA\")\n",
    "for index, value in enumerate(x):\n",
    "    plt.text(value-0.4,len(y_test_te[y_test_te==value]) , str(len(y_test_te[y_test_te==value])))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "r = \"షట్చక్రాలు మానవుని శరీరం లోని వెన్నుపూసలో ఉండే , దిగువ చెప్పిన ఆరు సూక్ష్మ స్థానాలను షట్చక్రాలు అంటారు* అష్ట భార్యలు* శ్రీకృష్ణుడి ఎనిమిది మంది భార్యలును అష్ట భార్యలు లేదా అష్టమహిషులు 1.రుక్మిణి,2. సత్యభామ,3. జాంబవతి,4. మిత్రవింద,5.భద్ర\"\n",
    "rv = vectorizer_te.transform([remove_punct_te(r)])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[7]\n"
     ]
    }
   ],
   "source": [
    "print(clf_dt_te.predict(rv))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### GUI ###"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "import PIL.Image,PIL.ImageTk\n",
    "def predict():\n",
    "    ans = remove_punct_te(str(textArea.get(\"1.0\",\"end\")))\n",
    "    X1 = vectorizer_te.transform([ans])\n",
    "    pred = clf_dt_te.predict(X1)\n",
    "                    \n",
    "    if(pred==0):\n",
    "        Label(gui, text =\"The predicted Tweet is of Class 0\").place(x=120,y=200)\n",
    "        photo0 = PhotoImage(file = r\"emoji_prediction_classPNGS\\\\class1.png\") \n",
    "        photoimage0 = photo0.subsample(1, 1) \n",
    "        Button(gui, image = photoimage0).place(x=150,y=250)\n",
    "        Button.pack()\n",
    "    if(pred==1):\n",
    "        Label(gui, text =\"The predicted Tweet is of Class 1\").place(x=120,y=200)\n",
    "        photo0 = PhotoImage(file = r\"emoji_prediction_classPNGS\\\\class2.png\") \n",
    "        photoimage0 = photo0.subsample(1, 1) \n",
    "        Button(gui, image = photoimage0).place(x=150,y=250)\n",
    "        Button.pack()\n",
    "    if(pred==2):\n",
    "        Label(gui, text =\"The predicted Tweet is of Class 2\").place(x=120,y=200)\n",
    "        photo0 = PhotoImage(file = r\"emoji_prediction_classPNGS\\\\class3.png\") \n",
    "        photoimage0 = photo0.subsample(1, 1) \n",
    "        Button(gui, image = photoimage0).place(x=150,y=250)\n",
    "        Button.pack()\n",
    "    if(pred==3):\n",
    "        Label(gui, text =\"The predicted Tweet is of Class 3\").place(x=120,y=200)\n",
    "        photo0 = PhotoImage(file = r\"emoji_prediction_classPNGS\\\\class4.png\") \n",
    "        photoimage0 = photo0.subsample(1, 1) \n",
    "        Button(gui, image = photoimage0).place(x=150,y=250)\n",
    "        Button.pack()\n",
    "    if(pred==4):\n",
    "        Label(gui, text =\"The predicted Tweet is of Class 4\").place(x=120,y=200)\n",
    "        photo0 = PhotoImage(file = r\"emoji_prediction_classPNGS\\\\class5.png\") \n",
    "        photoimage0 = photo0.subsample(1, 1) \n",
    "        Button(gui, image = photoimage0).place(x=150,y=250)\n",
    "        Button.pack()\n",
    "    if(pred==5):\n",
    "        Label(gui, text =\"The predicted Tweet is Class 5\").place(x=120,y=200)\n",
    "        photo0 = PhotoImage(file = r\"emoji_prediction_classPNGS\\\\class6.png\") \n",
    "        photoimage0 = photo0.subsample(1, 1) \n",
    "        Button(gui, image = photoimage0).place(x=150,y=350)\n",
    "        Button.pack()\n",
    "    if(pred==6):\n",
    "        Label(gui, text =\"The predicted Tweet is of Class 6\").place(x=120,y=200)\n",
    "        photo0 = PhotoImage(file = r\"emoji_prediction_classPNGS\\\\class7.png\") \n",
    "        photoimage0 = photo0.subsample(1, 1) \n",
    "        Button(gui, image = photoimage0).place(x=150,y=250)\n",
    "        Button.pack()\n",
    "    if(pred==7):\n",
    "        Label(gui, text =\"The predicted Tweet is of Class 7\").place(x=120,y=200)\n",
    "        photo0 = PhotoImage(file = r\"emoji_prediction_classPNGS\\\\class8.png\") \n",
    "        photoimage0 = photo0.subsample(1, 1) \n",
    "        Button(gui, image = photoimage0).place(x=150,y=250)\n",
    "        Button.pack()\n",
    "    if(pred==8):\n",
    "        Label(gui, text =\"The predicted Tweet is Class 8\").place(x=120,y=200)\n",
    "        photo0 = PhotoImage(file = r\"emoji_prediction_classPNGS\\\\class9.png\") \n",
    "        photoimage0 = photo0.subsample(1, 1) \n",
    "        Button(gui, image = photoimage0).place(x=150,y=250)\n",
    "        Button.pack()\n",
    "    if(pred==9):\n",
    "        Label(gui, text =\"The predicted Tweet is of Class 9\").place(x=120,y=200)\n",
    "        photo0 = PhotoImage(file = r\"emoji_prediction_classPNGS\\\\class10.png\") \n",
    "        photoimage0 = photo0.subsample(1, 1) \n",
    "        Button(gui, image = photoimage0).place(x=150,y=350)\n",
    "        Button.pack()\n",
    "    if(pred==10):\n",
    "        Label(gui, text =\"The predicted Tweet is Class 10\").place(x=120,y=200)\n",
    "        photo0 = PhotoImage(file = r\"emoji_prediction_classPNGS\\\\class11.png\") \n",
    "        photoimage0 = photo0.subsample(1, 1) \n",
    "        Button(gui, image = photoimage0).place(x=150,y=250)\n",
    "        Button.pack()\n",
    "    if(pred==11):\n",
    "        Label(gui, text =\"The predicted Tweet is of Class 11\").place(x=120,y=200)\n",
    "        photo0 = PhotoImage(file = r\"emoji_prediction_classPNGS\\\\class12.png\") \n",
    "        photoimage0 = photo0.subsample(1, 1) \n",
    "        Button(gui, image = photoimage0).place(x=150,y=350)\n",
    "        Button.pack()\n",
    "    if(pred==12):\n",
    "        Label(gui, text =\"The predicted Tweet is of Class 12\").place(x=120,y=200)\n",
    "        photo0 = PhotoImage(file = r\"emoji_prediction_classPNGS\\\\class13.png\") \n",
    "        photoimage0 = photo0.subsample(1, 1) \n",
    "        Button(gui, image = photoimage0).place(x=150,y=250)\n",
    "        Button.pack()\n",
    "    if(pred==13):\n",
    "        Label(gui, text =\"The predicted Tweet is of Class 13\").place(x=120,y=200)\n",
    "        photo0 = PhotoImage(file = r\"emoji_prediction_classPNGS\\\\class14.png\") \n",
    "        photoimage0 = photo0.subsample(1, 1) \n",
    "        Button(gui, image = photoimage0).place(x=150,y=350)\n",
    "        Button.pack()\n",
    "    if(pred==14):\n",
    "        Label(gui, text =\"The predicted Tweet is of Class 14\").place(x=120,y=200)\n",
    "        photo0 = PhotoImage(file = r\"emoji_prediction_classPNGS\\\\class15.png\") \n",
    "        photoimage0 = photo0.subsample(1, 1) \n",
    "        Button(gui, image = photoimage0).place(x=150,y=250)\n",
    "        Button.pack()\n",
    "    if(pred==15):\n",
    "        Label(gui, text =\"The predicted Tweet is of Class 15\").place(x=120,y=200)\n",
    "        photo0 = PhotoImage(file = r\"emoji_prediction_classPNGS\\\\class16.png\") \n",
    "        photoimage0 = photo0.subsample(1, 1) \n",
    "        Button(gui, image = photoimage0).place(x=150,y=250)\n",
    "        Button.pack()\n",
    "    if(pred==16):\n",
    "        Label(gui, text =\"The predicted Tweet is of Class 16\").place(x=120,y=200)\n",
    "        photo0 = PhotoImage(file = r\"emoji_prediction_classPNGS\\\\class17.png\") \n",
    "        photoimage0 = photo0.subsample(1, 1) \n",
    "        Button(gui, image = photoimage0).place(x=150,y=250)\n",
    "        Button.pack()\n",
    "    if(pred==17):\n",
    "        Label(gui, text =\"The predicted Tweet is of Class 17\").place(x=120,y=200)\n",
    "        photo0 = PhotoImage(file = r\"emoji_prediction_classPNGS\\\\class18.png\") \n",
    "        photoimage0 = photo0.subsample(1, 1) \n",
    "        Button(gui, image = photoimage0).place(x=150,y=250)\n",
    "        Button.pack()\n",
    "    if(pred==18):\n",
    "        Label(gui, text =\"The predicted Tweet is of Class 18\").place(x=120,y=200)\n",
    "        photo0 = PhotoImage(file = r\"emoji_prediction_classPNGS\\\\class19.png\") \n",
    "        photoimage0 = photo0.subsample(1, 1) \n",
    "        Button(gui, image = photoimage0).place(x=150,y=250)\n",
    "        Button.pack()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Exception in Tkinter callback\n",
      "Traceback (most recent call last):\n",
      "  File \"c:\\users\\kabilan n\\appdata\\local\\programs\\python\\python37\\lib\\tkinter\\__init__.py\", line 1705, in __call__\n",
      "    return self.func(*args)\n",
      "  File \"<ipython-input-27-7d38bcdc2699>\", line 15, in predict\n",
      "    photo0 = PhotoImage(file = r\"C:\\\\Users\\\\haris\\\\Downloads\\\\class2.png\")\n",
      "  File \"c:\\users\\kabilan n\\appdata\\local\\programs\\python\\python37\\lib\\tkinter\\__init__.py\", line 3545, in __init__\n",
      "    Image.__init__(self, 'photo', name, cnf, master, **kw)\n",
      "  File \"c:\\users\\kabilan n\\appdata\\local\\programs\\python\\python37\\lib\\tkinter\\__init__.py\", line 3501, in __init__\n",
      "    self.tk.call(('image', 'create', imgtype, name,) + options)\n",
      "_tkinter.TclError: couldn't open \"C:\\\\Users\\\\haris\\\\Downloads\\\\class2.png\": no such file or directory\n"
     ]
    }
   ],
   "source": [
    "from matplotlib.figure import Figure \n",
    "from matplotlib.backends.backend_tkagg import (FigureCanvasTkAgg,  \n",
    "NavigationToolbar2Tk)\n",
    "import tkinter as tk\n",
    "from pandas import DataFrame\n",
    "import matplotlib.pyplot as plt\n",
    "from matplotlib.backends.backend_tkagg import FigureCanvasTkAgg\n",
    "import tkinter as tk\n",
    "from tkinter import *\n",
    "from matplotlib.figure import Figure \n",
    "gui = Tk()\n",
    "gui.title(\"Emoji Prediction - Report\")\n",
    "gui.geometry('500x400') \n",
    "gui.configure(bg='steel blue')\n",
    "Label(gui, text = \"Enter Tweets : \").place(x=20,y=160)\n",
    "textArea = Text(gui, height = 2, width = 25, font = \"cambria 13\")\n",
    "textArea.place(x=100,y=150)\n",
    "run_button = Button(gui,  command = predict, text = \"predict\")\n",
    "run_button.pack() \n",
    "run_button.place(x=340,y=155)\n",
    "label = Label(gui)\n",
    "label.pack()\n",
    "mainloop()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Deep learning Model ###"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = Sequential()\n",
    "model.add(Dense(500, activation='relu', input_dim = 3732))\n",
    "model.add(Dense(100, activation='relu'))\n",
    "model.add(Dense(50, activation='relu'))\n",
    "model.add(Dense(19, activation='softmax'))\n",
    "\n",
    "# Compile the model\n",
    "model.compile(optimizer='adam', \n",
    "              loss='categorical_crossentropy', \n",
    "              metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 13592 samples\n",
      "Epoch 1/20\n",
      "13592/13592 [==============================] - 3s 256us/sample - loss: 0.1012 - accuracy: 0.9439\n",
      "Epoch 2/20\n",
      "13592/13592 [==============================] - 3s 255us/sample - loss: 0.1064 - accuracy: 0.9465\n",
      "Epoch 3/20\n",
      "13592/13592 [==============================] - 3s 256us/sample - loss: 0.1010 - accuracy: 0.9470\n",
      "Epoch 4/20\n",
      "13592/13592 [==============================] - 4s 258us/sample - loss: 0.1020 - accuracy: 0.9462\n",
      "Epoch 5/20\n",
      "13592/13592 [==============================] - 3s 256us/sample - loss: 0.1038 - accuracy: 0.9452\n",
      "Epoch 6/20\n",
      "13592/13592 [==============================] - 4s 260us/sample - loss: 0.1056 - accuracy: 0.9470\n",
      "Epoch 7/20\n",
      "13592/13592 [==============================] - 4s 258us/sample - loss: 0.1173 - accuracy: 0.9429\n",
      "Epoch 8/20\n",
      "13592/13592 [==============================] - 4s 259us/sample - loss: 0.1328 - accuracy: 0.9411\n",
      "Epoch 9/20\n",
      "13592/13592 [==============================] - 4s 259us/sample - loss: 0.1141 - accuracy: 0.9447\n",
      "Epoch 10/20\n",
      "13592/13592 [==============================] - 4s 259us/sample - loss: 0.1027 - accuracy: 0.9460\n",
      "Epoch 11/20\n",
      "13592/13592 [==============================] - 4s 258us/sample - loss: 0.0994 - accuracy: 0.9485\n",
      "Epoch 12/20\n",
      "13592/13592 [==============================] - 4s 261us/sample - loss: 0.0995 - accuracy: 0.9461\n",
      "Epoch 13/20\n",
      "13592/13592 [==============================] - 4s 265us/sample - loss: 0.0993 - accuracy: 0.9463\n",
      "Epoch 14/20\n",
      "13592/13592 [==============================] - 4s 260us/sample - loss: 0.0996 - accuracy: 0.9473\n",
      "Epoch 15/20\n",
      "13592/13592 [==============================] - 4s 262us/sample - loss: 0.1006 - accuracy: 0.9489\n",
      "Epoch 16/20\n",
      "13592/13592 [==============================] - 4s 263us/sample - loss: 0.1017 - accuracy: 0.9465\n",
      "Epoch 17/20\n",
      "13592/13592 [==============================] - 4s 261us/sample - loss: 0.1054 - accuracy: 0.9460\n",
      "Epoch 18/20\n",
      "13592/13592 [==============================] - 4s 261us/sample - loss: 0.1094 - accuracy: 0.9439\n",
      "Epoch 19/20\n",
      "13592/13592 [==============================] - 4s 262us/sample - loss: 0.1056 - accuracy: 0.9472\n",
      "Epoch 20/20\n",
      "13592/13592 [==============================] - 4s 264us/sample - loss: 0.1063 - accuracy: 0.9452\n",
      "\n",
      " 70.65066123008728 secs\n"
     ]
    }
   ],
   "source": [
    "start=time.time()\n",
    "model.fit(X_train_d, y_train_d, epochs = 20)\n",
    "end=time.time()\n",
    "tt=end-start\n",
    "print(\"Time taken to train the sequence :\")\n",
    "print(tt,\"secs\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy on training data: 0.9489405751228333 \n",
      " Error on training data: 0.05105942487716675\n",
      "Accuracy on test data: 0.5361977815628052 \n",
      " Error on test data: 0.4638022184371948\n"
     ]
    }
   ],
   "source": [
    "pred_train= model.predict(X_train_d)\n",
    "scores = model.evaluate(X_train_d, y_train_d, verbose=0)\n",
    "print('Accuracy on training data: {} \\n Error on training data: {}'.format(scores[1], 1 - scores[1]))   \n",
    " \n",
    "pred_test= model.predict(X_test_d)\n",
    "scores2 = model.evaluate(X_test_d, y_test_d, verbose=0)\n",
    "print('Accuracy on test data: {} \\n Error on test data: {}'.format(scores2[1], 1 - scores2[1])) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Saved model to disk\n"
     ]
    }
   ],
   "source": [
    "model_json = model.to_json()\n",
    "with open(\"model.json\", \"w\") as json_file:\n",
    "    json_file.write(model_json)\n",
    "# serialize weights to HDF5\n",
    "model.save_weights(\"model.h5\")\n",
    "print(\"Saved model to disk\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "from tensorflow.keras.models import model_from_json"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loaded model from disk\n",
      "accuracy: 95.40%\n"
     ]
    }
   ],
   "source": [
    "json_file = open('model.json', 'r')\n",
    "loaded_model_json = json_file.read()\n",
    "json_file.close()\n",
    "loaded_model = model_from_json(loaded_model_json)\n",
    "# load weights into new model\n",
    "loaded_model.load_weights(\"model.h5\")\n",
    "print(\"Loaded model from disk\")\n",
    " \n",
    "# evaluate loaded model on test data\n",
    "loaded_model.compile(loss='binary_crossentropy', optimizer='rmsprop', metrics=['accuracy'])\n",
    "score = loaded_model.evaluate(X_test_d,y_test_d,verbose=0)\n",
    "print(\"%s: %.2f%%\" % (loaded_model.metrics_names[1], score[1]*100))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [],
   "source": [
    "l = \"దొంగల స్థావరాలు ఎవరికి తెలియకూడదని దొంగలు జాగ్రత్తపడతారు,సమయం ఆసన్నమైంది నువ్వు నిప్పువో లేదా కంపువో నిరూపించుకో గజదొంగ …\"\n",
    "l1 = vectorizer_d.transform([remove_punct(r)]).toarray()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1, 3732)"
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "l1.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "11\n"
     ]
    }
   ],
   "source": [
    "print(np.argmax(loaded_model.predict(l1)))"
   ]
  }
 ],
 "metadata": {
  "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": "3.9.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}