mainNotebook.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Twitter Disaster Classification: Natural Language Processing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Importing Libraries\n",
    "\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from collections import defaultdict\n",
    "from sklearn.pipeline import Pipeline\n",
    "from sklearn.compose import ColumnTransformer\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import accuracy_score, precision_score, recall_score, f1_score\n",
    "from sklearn.feature_extraction.text import CountVectorizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Importing data\n",
    "df = pd.read_csv('data/train.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Feature Selection\n",
    "X = df['text']\n",
    "y = df['target']\n",
    "\n",
    "#vectorizer = CountVectorizer(ngram_range=(2,2))\n",
    "#X = vectorizer.fit_transform(X)\n",
    "#X = pd.DataFrame(X.toarray(), columns=vectorizer.get_feature_names())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Let's split the data for training, cross-validation, and testing\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.2,\n",
    "                                                    random_state = 72);\n",
    "X_train, X_val, y_train, y_val = train_test_split(X_train, y_train, \n",
    "                                                  test_size = 0.2, \n",
    "                                                  random_state = 72);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Plotting:\n",
    "acc = []\n",
    "pre = []\n",
    "rec=[]\n",
    "fs=[]\n",
    "def evaluation(y_val, predictions):\n",
    "    \"\"\" Prints out evaluation information.\n",
    "    \n",
    "        y_val: true value of the targets\n",
    "        predictions: predicted value of the targets\"\"\"\n",
    "    # Let's look at the evaluation metrics:\n",
    "    ## 1. Accuracy\n",
    "    accuracy = accuracy_score(y_val, predictions)\n",
    "    acc.append(accuracy)\n",
    "    \n",
    "    ## 2. Precision\n",
    "    precision = precision_score(y_val, predictions)\n",
    "    pre.append(precision)\n",
    "    \n",
    "    ## 3. Recall\n",
    "    recall = recall_score(y_val, predictions)\n",
    "    rec.append(recall)\n",
    "    \n",
    "    ## 4. F1 score\n",
    "    f1 = f1_score(y_val, predictions)\n",
    "    fs.append(f1)\n",
    "    print(\"Accuracy: \",accuracy, \"\\nPrecision:\", \n",
    "          precision, \"\\nRecall:   \",recall, \"\\nF1:       \",f1)\n",
    "    return f1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy:  0.7996715927750411 \n",
      "Precision: 0.8171806167400881 \n",
      "Recall:    0.6973684210526315 \n",
      "F1:        0.7525354969574035\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.7525354969574035"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Algorithm 1: Naive Bayes\n",
    "\n",
    "from sklearn.naive_bayes import MultinomialNB\n",
    "classifier = MultinomialNB()\n",
    "vectorizer = CountVectorizer(ngram_range=(1,2))\n",
    "\n",
    "# Make a pipeline\n",
    "stepsNB = [(\"count\", vectorizer),(\"classifier\", classifier)]\n",
    "pipeNB = Pipeline(steps = stepsNB)\n",
    "\n",
    "# Fit the pipe:\n",
    "pipeNB.fit(X_train, y_train)\n",
    "predictionsNB = pipeNB.predict(X_val)\n",
    "\n",
    "# Evaluation\n",
    "evaluation(y_val, predictionsNB)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy:  0.7750410509031199 \n",
      "Precision: 0.776824034334764 \n",
      "Recall:    0.6804511278195489 \n",
      "F1:        0.7254509018036073\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.7254509018036073"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Algorithm 2: AdaBoost\n",
    "from sklearn.ensemble import AdaBoostClassifier\n",
    "modelAB = AdaBoostClassifier(n_estimators = 150, random_state = 1)\n",
    "stepsAB = [(\"count\", vectorizer),(\"adaboost\", modelAB)]\n",
    "pipeAB = Pipeline(steps = stepsAB)\n",
    "pipeAB.fit(X_train, y_train)\n",
    "predictionsAB = pipeAB.predict(X_val)\n",
    "evaluation(y_val, predictionsAB)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy:  0.7947454844006568 \n",
      "Precision: 0.827906976744186 \n",
      "Recall:    0.6691729323308271 \n",
      "F1:        0.74012474012474\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.74012474012474"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Algorithm 3: Logistic Regression\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "modelLR = LogisticRegression(random_state = 1, max_iter = 100000)\n",
    "stepsLR = [(\"counts\",vectorizer), (\"LR\", modelLR)]\n",
    "pipeLR = Pipeline(steps = stepsLR)\n",
    "pipeLR.fit(X_train, y_train)\n",
    "predictionsLR = pipeLR.predict(X_val)\n",
    "evaluation(y_val, predictionsLR)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy:  0.7848932676518884 \n",
      "Precision: 0.8609625668449198 \n",
      "Recall:    0.6052631578947368 \n",
      "F1:        0.7108167770419427\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.7108167770419427"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Algorithm 4: Support Vector Classifier\n",
    "from sklearn.svm import SVC\n",
    "modelSVC = SVC(C = 2, degree = 1)\n",
    "stepsSVC = [(\"counts\",vectorizer), (\"SVC\", modelSVC)]\n",
    "pipeSVC = Pipeline(steps = stepsSVC)\n",
    "pipeSVC.fit(X_train, y_train)\n",
    "predictionsSVC = pipeSVC.predict(X_val)\n",
    "evaluation(y_val, predictionsSVC)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy:  0.7635467980295566 \n",
      "Precision: 0.7811059907834101 \n",
      "Recall:    0.6372180451127819 \n",
      "F1:        0.7018633540372671\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.7018633540372671"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Algorithm 5: XGBoost\n",
    "from xgboost import XGBClassifier\n",
    "modelXGB = XGBClassifier(max_depth = 100, learning_rate = .1)\n",
    "stepsXGB = [(\"counts\",vectorizer), (\"XGB\", modelXGB)]\n",
    "pipeXGB = Pipeline(steps = stepsXGB)\n",
    "pipeXGB.fit(X_train, y_train)\n",
    "predictionsXGB = pipeXGB.predict(X_val)\n",
    "evaluation(y_val, predictionsXGB)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Plot 1\n",
    "\n",
    "fig = plt.figure();\n",
    "plt.plot([0, 10, 20, 30, 40],acc,'ko-')\n",
    "plt.xticks([0, 10, 20, 30, 40], [\"Naive-Bayes\", \"AdaBoost\",\n",
    "                                 \"Logistic Regression\", \"SVM\", \"XGBoost\"])\n",
    "plt.plot([0, 10, 20, 30, 40],pre,'m*-')\n",
    "plt.plot([0, 10, 20, 30, 40],rec,'rs-')\n",
    "plt.plot([0, 10, 20, 30, 40],fs, 'g>-')\n",
    "plt.legend([\"Accuracy\",\"Precision\",\"Recall\",\"F1 score\"])\n",
    "plt.savefig(fname = \"comparison.png\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Plot 2\n",
    "\n",
    "# We can take a look at the importance of tokens with \n",
    "# the help of Logistic Regression Model\n",
    "coeffs = modelLR.coef_\n",
    "df = pd.read_csv('data/train.csv')\n",
    "X = df['text']\n",
    "y = df['target']\n",
    "vectorizer = CountVectorizer(ngram_range=(1,2))\n",
    "Xa = vectorizer.fit_transform(X)\n",
    "X = pd.DataFrame(Xa.toarray(), columns=vectorizer.get_feature_names())\n",
    "# Since this is a very high dimensional space, we will look at the first few tokens\n",
    "\n",
    "x_axis = np.linspace(2,500, 7)\n",
    "plt.plot(x_axis, coeffs[0][0:7])\n",
    "plt.xticks(x_axis, [\"\\\"00\\\"\", \"\\\"00 11\\\"\", \"\\\"00 25\\\"\", \"\\\"00 52\\\"\",\n",
    "                    \"\\\"00 am\\\"\", \"\\\"00 at\\\"\", \"\\\"00 bestseller\\\"\"])\n",
    "plt.xlabel(\"TOKENS\")\n",
    "plt.xlim([-2, 500])\n",
    "plt.ylabel(\"WEIGHT OF TOKEN\")\n",
    "plt.title(\"WEIGHT OF THE FIRST FEW TOKENS\")\n",
    "plt.savefig(fname = \"weight.png\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy:  0.799080761654629 \n",
      "Precision: 0.823943661971831 \n",
      "Recall:    0.6943620178041543 \n",
      "F1:        0.7536231884057971\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.7536231884057971"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Final Results\n",
    "# Test Set evaluation\n",
    "\n",
    "testPredictions = pipeNB.predict(X_test)\n",
    "evaluation(y_test, testPredictions)"
   ]
  }
 ],
 "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": "3.6.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}