NLP_supervised_classify_topic_of_documents.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# classify documents on 20 topics\n",
    "## via scikit-learn\n",
    "1. raw data are news from 20 topics\n",
    "2. features extracted from documents include count of words, term of frequency (tf), and term frequency times inverse document frequency (tf-idf)\n",
    "3. classifers include Naive Bayes, Support Vector Machine, and Random Forest\n",
    "4. classifer performance is measured by precision, recall, and F1-score\n",
    "\n",
    "## note:\n",
    "1. counts must be the input to tf/tfidf"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [],
   "source": [
    "import warnings\n",
    "warnings.filterwarnings('ignore')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [],
   "source": [
    "# get data\n",
    "from sklearn.datasets import fetch_20newsgroups\n",
    "# get functions used for pre-processing raw data\n",
    "from sklearn.feature_extraction.text import CountVectorizer # for count of words\n",
    "from sklearn.feature_extraction.text import TfidfTransformer # for tf or tf-idf\n",
    "# get classifer\n",
    "from sklearn.naive_bayes import MultinomialNB # Naive Bayes\n",
    "from sklearn.linear_model import SGDClassifier # SVM\n",
    "from sklearn.ensemble import RandomForestClassifier # random forest\n",
    "# get measurement\n",
    "from sklearn import metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [],
   "source": [
    "## 1. get all data, and list 20 topics\n",
    "twenty_train = fetch_20newsgroups(subset='train', shuffle=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "11314"
      ]
     },
     "execution_count": 62,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(twenty_train['data'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['alt.atheism',\n",
       " 'comp.graphics',\n",
       " 'comp.os.ms-windows.misc',\n",
       " 'comp.sys.ibm.pc.hardware',\n",
       " 'comp.sys.mac.hardware',\n",
       " 'comp.windows.x',\n",
       " 'misc.forsale',\n",
       " 'rec.autos',\n",
       " 'rec.motorcycles',\n",
       " 'rec.sport.baseball',\n",
       " 'rec.sport.hockey',\n",
       " 'sci.crypt',\n",
       " 'sci.electronics',\n",
       " 'sci.med',\n",
       " 'sci.space',\n",
       " 'soc.religion.christian',\n",
       " 'talk.politics.guns',\n",
       " 'talk.politics.mideast',\n",
       " 'talk.politics.misc',\n",
       " 'talk.religion.misc']"
      ]
     },
     "execution_count": 63,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "twenty_train.target_names"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "\"From: lerxst@wam.umd.edu (where's my thing)\\nSubject: WHAT car is this!?\\nNntp-Posting-Host: rac3.wam.umd.edu\\nOrganization: University of Maryland, College Park\\nLines: 15\\n\\n I was wondering if anyone out there could enlighten me on this car I saw\\nthe other day. It was a 2-door sports car, looked to be from the late 60s/\\nearly 70s. It was called a Bricklin. The doors were really small. In addition,\\nthe front bumper was separate from the rest of the body. This is \\nall I know. If anyone can tellme a model name, engine specs, years\\nof production, where this car is made, history, or whatever info you\\nhave on this funky looking car, please e-mail.\\n\\nThanks,\\n- IL\\n   ---- brought to you by your neighborhood Lerxst ----\\n\\n\\n\\n\\n\""
      ]
     },
     "execution_count": 64,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# one sample of the text\n",
    "twenty_train.data[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'rec.autos'"
      ]
     },
     "execution_count": 65,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# the topic of this text\n",
    "twenty_train.target_names[twenty_train.target[0]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19}"
      ]
     },
     "execution_count": 66,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "set(twenty_train.target)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [],
   "source": [
    "## 2. pre-process raw text\n",
    "### count of words / bag of words\n",
    "count_vect = CountVectorizer()\n",
    "X_train_counts = count_vect.fit_transform(twenty_train.data)\n",
    "### term of frequency\n",
    "tf_transformer = TfidfTransformer(use_idf=False).fit(X_train_counts)\n",
    "X_train_tf = tf_transformer.transform(X_train_counts)\n",
    "### term frequency times inverse document frequency\n",
    "tfidf_transformer = TfidfTransformer(use_idf=True).fit(X_train_counts)\n",
    "X_train_tfidf = tfidf_transformer.transform(X_train_counts)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## A. Navie Bayes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [],
   "source": [
    "## 3. deploy classifers\n",
    "### NB\n",
    "clf_NB = MultinomialNB().fit(X_train_tfidf, twenty_train.target) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "'God is love' => soc.religion.christian\n",
      "'OpenGL on the GPU is fast' => rec.autos\n"
     ]
    }
   ],
   "source": [
    "## 4. example:\n",
    "##           what is the topic of these two new documents\n",
    "##           when classifier is navie bayers and the feature is tfidf\n",
    "docs_new = ['God is love', 'OpenGL on the GPU is fast']\n",
    "X_new_counts = count_vect.transform(docs_new)\n",
    "X_new_tfidf = tfidf_transformer.transform(X_new_counts)\n",
    "\n",
    "ex_predicted = clf_NB.predict(X_new_tfidf)\n",
    "\n",
    "for doc, category in zip(docs_new, ex_predicted):\n",
    "    print('%r => %s' % (doc, twenty_train.target_names[category]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "                          precision    recall  f1-score   support\n",
      "\n",
      "             alt.atheism       0.80      0.52      0.63       319\n",
      "           comp.graphics       0.81      0.65      0.72       389\n",
      " comp.os.ms-windows.misc       0.82      0.65      0.73       394\n",
      "comp.sys.ibm.pc.hardware       0.67      0.78      0.72       392\n",
      "   comp.sys.mac.hardware       0.86      0.77      0.81       385\n",
      "          comp.windows.x       0.89      0.75      0.82       395\n",
      "            misc.forsale       0.93      0.69      0.80       390\n",
      "               rec.autos       0.85      0.92      0.88       396\n",
      "         rec.motorcycles       0.94      0.93      0.93       398\n",
      "      rec.sport.baseball       0.92      0.90      0.91       397\n",
      "        rec.sport.hockey       0.89      0.97      0.93       399\n",
      "               sci.crypt       0.59      0.97      0.74       396\n",
      "         sci.electronics       0.84      0.60      0.70       393\n",
      "                 sci.med       0.92      0.74      0.82       396\n",
      "               sci.space       0.84      0.89      0.87       394\n",
      "  soc.religion.christian       0.44      0.98      0.61       398\n",
      "      talk.politics.guns       0.64      0.94      0.76       364\n",
      "   talk.politics.mideast       0.93      0.91      0.92       376\n",
      "      talk.politics.misc       0.96      0.42      0.58       310\n",
      "      talk.religion.misc       0.97      0.14      0.24       251\n",
      "\n",
      "                accuracy                           0.77      7532\n",
      "               macro avg       0.83      0.76      0.76      7532\n",
      "            weighted avg       0.82      0.77      0.77      7532\n",
      "\n"
     ]
    }
   ],
   "source": [
    "## 5. measure the performance of the selected classifier\n",
    "twenty_test = fetch_20newsgroups(subset='test', shuffle=True, random_state=42)\n",
    "twenty_test_counts = count_vect.transform(twenty_test.data)\n",
    "twenty_test_tfidf = tfidf_transformer.transform(twenty_test_counts)\n",
    "\n",
    "predicted = clf_NB.predict(twenty_test_tfidf)\n",
    "\n",
    "print(metrics.classification_report(twenty_test.target, predicted, target_names=twenty_test.target_names))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [],
   "source": [
    "#metrics.confusion_matrix(twenty_test.target, predicted)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## B. Support Vector Machine"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "                          precision    recall  f1-score   support\n",
      "\n",
      "             alt.atheism       0.73      0.71      0.72       319\n",
      "           comp.graphics       0.78      0.72      0.75       389\n",
      " comp.os.ms-windows.misc       0.73      0.78      0.75       394\n",
      "comp.sys.ibm.pc.hardware       0.74      0.67      0.70       392\n",
      "   comp.sys.mac.hardware       0.81      0.83      0.82       385\n",
      "          comp.windows.x       0.84      0.76      0.80       395\n",
      "            misc.forsale       0.84      0.90      0.87       390\n",
      "               rec.autos       0.91      0.90      0.90       396\n",
      "         rec.motorcycles       0.93      0.96      0.95       398\n",
      "      rec.sport.baseball       0.88      0.90      0.89       397\n",
      "        rec.sport.hockey       0.88      0.99      0.93       399\n",
      "               sci.crypt       0.84      0.96      0.90       396\n",
      "         sci.electronics       0.83      0.62      0.71       393\n",
      "                 sci.med       0.87      0.86      0.87       396\n",
      "               sci.space       0.84      0.96      0.90       394\n",
      "  soc.religion.christian       0.76      0.94      0.84       398\n",
      "      talk.politics.guns       0.70      0.92      0.80       364\n",
      "   talk.politics.mideast       0.90      0.93      0.92       376\n",
      "      talk.politics.misc       0.89      0.55      0.68       310\n",
      "      talk.religion.misc       0.85      0.41      0.55       251\n",
      "\n",
      "                accuracy                           0.82      7532\n",
      "               macro avg       0.83      0.81      0.81      7532\n",
      "            weighted avg       0.83      0.82      0.82      7532\n",
      "\n"
     ]
    }
   ],
   "source": [
    "\n",
    "clf_SVM = SGDClassifier(loss='hinge', penalty='l2',\n",
    "                           alpha=1e-3, random_state=42,\n",
    "                           max_iter=5, tol=None).fit(X_train_tfidf, twenty_train.target)\n",
    "\n",
    "predicted_svm = clf_SVM.predict(twenty_test_tfidf)\n",
    "\n",
    "print(metrics.classification_report(twenty_test.target, predicted_svm, target_names=twenty_test.target_names))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## C. Random Forest"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "                          precision    recall  f1-score   support\n",
      "\n",
      "             alt.atheism       0.71      0.62      0.66       319\n",
      "           comp.graphics       0.58      0.70      0.64       389\n",
      " comp.os.ms-windows.misc       0.63      0.74      0.68       394\n",
      "comp.sys.ibm.pc.hardware       0.66      0.65      0.65       392\n",
      "   comp.sys.mac.hardware       0.70      0.78      0.74       385\n",
      "          comp.windows.x       0.74      0.69      0.71       395\n",
      "            misc.forsale       0.75      0.89      0.81       390\n",
      "               rec.autos       0.81      0.79      0.80       396\n",
      "         rec.motorcycles       0.89      0.90      0.90       398\n",
      "      rec.sport.baseball       0.80      0.90      0.85       397\n",
      "        rec.sport.hockey       0.89      0.92      0.91       399\n",
      "               sci.crypt       0.86      0.91      0.89       396\n",
      "         sci.electronics       0.66      0.46      0.54       393\n",
      "                 sci.med       0.83      0.68      0.75       396\n",
      "               sci.space       0.82      0.86      0.84       394\n",
      "  soc.religion.christian       0.68      0.91      0.78       398\n",
      "      talk.politics.guns       0.66      0.86      0.75       364\n",
      "   talk.politics.mideast       0.95      0.78      0.86       376\n",
      "      talk.politics.misc       0.85      0.49      0.62       310\n",
      "      talk.religion.misc       0.74      0.31      0.43       251\n",
      "\n",
      "                accuracy                           0.75      7532\n",
      "               macro avg       0.76      0.74      0.74      7532\n",
      "            weighted avg       0.76      0.75      0.75      7532\n",
      "\n"
     ]
    }
   ],
   "source": [
    "rf = RandomForestClassifier().fit(X_train_tfidf, twenty_train.target)\n",
    "predicted_rf = rf.predict(twenty_test_tfidf)\n",
    "\n",
    "print(metrics.classification_report(twenty_test.target, predicted_rf, target_names=twenty_test.target_names))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}