adminsolvve · AlexGnedenko · May 20, 2019
diff --git a/Naive_Bayes_text_classification/Instructions.pdf b/Naive_Bayes_text_classification/Instructions.pdf
diff --git a/Naive_Bayes_text_classification/NaiveBayesTextClassification.ipynb b/Naive_Bayes_text_classification/NaiveBayesTextClassification.ipynb
@@ -0,0 +1,358 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "# Naive Bayes Text Classification"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Using libraries:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 10,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import csv\n",
+    "import math\n",
+    "import nltk"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Global variable:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 11,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "dataBaseFile = \"traning.csv\"\n",
+    "dataBase = None"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Support method printDataBase:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 12,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def printDataBase():\n",
+    "    for i in dataBase:\n",
+    "        print(i)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### Data preprocesing:"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Reading .csv file:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 13,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def readCSVFileToArray(dataBase):\n",
+    "    with open(dataBaseFile, 'r') as fp:\n",
+    "        reader = csv.reader(fp, delimiter=',', quotechar='\"')\n",
+    "        dataBase = [row for row in reader]\n",
+    "        fp.close()    \n",
+    "    for i in range(0,len(dataBase)):\n",
+    "        temp = dataBase[i][-1]\n",
+    "        dataBase[i][-1] = dataBase[i][0]\n",
+    "        dataBase[i][0] = temp\n",
+    "    return dataBase"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Spliting:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 14,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def splitDataBase(dataBase):\n",
+    "    for i in range(0,len(dataBase)):        \n",
+    "        for j in range(1,len(dataBase[i])):\n",
+    "            tempList = str(dataBase[i][1]).split()\n",
+    "            dataBase[i].pop(1)\n",
+    "            for k in range(0,len(tempList)):\n",
+    "                dataBase[i].append(tempList[k])\n",
+    "    return dataBase"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Lowercase:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 15,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def lowercase(dataBase):\n",
+    "    for i in range(0,len(dataBase)):\n",
+    "        for j in range(1,len(dataBase[i])):\n",
+    "            dataBase[i][j] = dataBase[i][j].lower()\n",
+    "    return dataBase"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Remove STOP-WORDS:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 16,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def deleteSTOPWORDS(dataBase):\n",
+    "    nltk.download('stopwords')\n",
+    "    for i in range(0,len(dataBase)):\n",
+    "        newData = []\n",
+    "        for j in dataBase[i]:\n",
+    "            if (j not in nltk.corpus.stopwords.words('english')):\n",
+    "                newData.append(j)\n",
+    "        dataBase[i] =  newData\n",
+    "    return dataBase"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### Naive Bayes:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 17,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "class NaiveBayes:\n",
+    "    dataBase = None\n",
+    "    classesProb = []\n",
+    "    probOccurrence = []\n",
+    "    def __init__(self, dataBase):\n",
+    "        self.dataBase = dataBase\n",
+    "            \n",
+    "    def countClassesProb(self):\n",
+    "        for doc in self.dataBase:\n",
+    "            if (doc[0] not in self.classesProb):\n",
+    "                self.classesProb.append(doc[0])\n",
+    "        for i in range(0,len(self.classesProb)):\n",
+    "            self.classesProb[i] = 0\n",
+    "        for doc in self.dataBase:\n",
+    "            self.classesProb[int(doc[0])] = int(self.classesProb[int(doc[0])]) + 1        \n",
+    "        for i in range(0,len(self.classesProb)):\n",
+    "            self.classesProb[i] = self.classesProb[i] / len(self.dataBase)      \n",
+    "            \n",
+    "    def countProbOccurrenceInClasses(self, sample):\n",
+    "        self.countClassesProb()\n",
+    "        tableVocabulary = [[]]\n",
+    "        \n",
+    "        #generating tableVocabulary\n",
+    "        for i in range(0,len(self.dataBase)):\n",
+    "            for j in range(1,len(self.dataBase[i])):\n",
+    "                if (self.dataBase[i][j] not in tableVocabulary[0]):\n",
+    "                    tableVocabulary[0].append(self.dataBase[i][j])\n",
+    "        tableVocabulary[0].append(\"ALL\")\n",
+    "        \n",
+    "        for i in range(0,len(self.classesProb)):\n",
+    "            tableVocabulary.append([])\n",
+    "        \n",
+    "        for k in range(1,len(tableVocabulary)):\n",
+    "            for q in range(0,len(tableVocabulary[0])):\n",
+    "                tableVocabulary[k].append(0)\n",
+    "        \n",
+    "        for i in range(0,len(self.dataBase)):\n",
+    "            for j in range(1,len(self.dataBase[i])):\n",
+    "                for k in range(0,len(tableVocabulary[0])):\n",
+    "                    if (tableVocabulary[0][k]==self.dataBase[i][j]):\n",
+    "                        tableVocabulary[int(dataBase[i][0])+1][k] += 1\n",
+    "                        \n",
+    "        print(tableVocabulary)\n",
+    "                        \n",
+    "        for i in range(1,len(tableVocabulary)):\n",
+    "            for j in range(0,len(tableVocabulary[i])-1):\n",
+    "                tableVocabulary[i][j] += 1\n",
+    "                \n",
+    "        for i in range(1,len(tableVocabulary)):\n",
+    "            for j in range(0,len(tableVocabulary[i])-1):\n",
+    "                tableVocabulary[i][len(tableVocabulary[i])-1] += tableVocabulary[i][j]\n",
+    "                        \n",
+    "        for i in range(1,len(tableVocabulary)):\n",
+    "            for j in range(0,len(tableVocabulary[i])-1):\n",
+    "                tableVocabulary[i][j] /= tableVocabulary[i][len(tableVocabulary[i])-1]  \n",
+    "        \n",
+    "        #preprosesing sample to correct x-vector\n",
+    "        x = [[]]\n",
+    "        x[0].append('0')\n",
+    "        x[0].append(sample)\n",
+    "        x = lowercase(x)\n",
+    "        x = splitDataBase(x)\n",
+    "        x = deleteSTOPWORDS(x)\n",
+    "        \n",
+    "        #countProbOccurrenceInClasses        \n",
+    "        for c in range(0,len(self.classesProb)):\n",
+    "            self.probOccurrence.append(0)\n",
+    "            P = 0\n",
+    "            for i in range(1,len(x[0])):\n",
+    "                for j in range(0,len(tableVocabulary[0])-1):\n",
+    "                    if (x[0][i] == tableVocabulary[0][j]):\n",
+    "                        P = P + math.log(tableVocabulary[c+1][j])\n",
+    "            self.probOccurrence[c] = math.log(self.classesProb[c]) + P            \n",
+    "        \n",
+    "    def fit(self):        \n",
+    "        accuracy = 0\n",
+    "        trainDataBase = []\n",
+    "        testDataBase = []\n",
+    "        for i in range(0, len(self.dataBase)):\n",
+    "            if ((i+1)%5 == 0):\n",
+    "                testDataBase.append(self.dataBase[i])\n",
+    "            else:\n",
+    "                trainDataBase.append(self.dataBase[i])\n",
+    "        self.dataBase = trainDataBase;\n",
+    "        for i in range(0, len(testDataBase)):\n",
+    "            sample = \"\"\n",
+    "            for j in range(1, len(testDataBase[i])):\n",
+    "                sample = sample + \" \" + testDataBase[i][j]\n",
+    "            if (self.predict(sample)[1]==testDataBase[i][0]):\n",
+    "                accuracy += 1\n",
+    "        accuracy = (accuracy*100)/len(testDataBase)\n",
+    "        accuracy = str(int(accuracy)) + \"%\"\n",
+    "        return accuracy  \n",
+    "    \n",
+    "    def predict(self, sample):\n",
+    "        answer = [sample]\n",
+    "        answer.append(str(0))\n",
+    "        self.countProbOccurrenceInClasses(sample)\n",
+    "        m = self.probOccurrence[0]\n",
+    "        for i in range(0,len(self.probOccurrence)):\n",
+    "            if (self.probOccurrence[i]>m):\n",
+    "                m = self.probOccurrence[i]\n",
+    "                answer[-1] = str(i)\n",
+    "        return answer"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### Test model:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 18,
+   "metadata": {
+    "scrolled": true
+   },
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "[['chinese', 'beijing', 'shanghai', 'macao', 'tokyo', 'japan', 'ALL'], [5, 1, 1, 1, 0, 0, 0], [1, 0, 0, 0, 1, 1, 0]]\n"
+     ]
+    },
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "[nltk_data] Downloading package stopwords to\n",
+      "[nltk_data]     C:\\Users\\Александр\\AppData\\Roaming\\nltk_data...\n",
+      "[nltk_data]   Package stopwords is already up-to-date!\n",
+      "[nltk_data] Downloading package stopwords to\n",
+      "[nltk_data]     C:\\Users\\Александр\\AppData\\Roaming\\nltk_data...\n",
+      "[nltk_data]   Package stopwords is already up-to-date!\n"
+     ]
+    },
+    {
+     "data": {
+      "text/plain": [
+       "['Chinese The Chinese In Chinese Tokyo The In The Japan', '0']"
+      ]
+     },
+     "execution_count": 18,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "dataBase = readCSVFileToArray(dataBase)\n",
+    "dataBase = lowercase(dataBase)\n",
+    "dataBase = splitDataBase(dataBase)\n",
+    "dataBase = deleteSTOPWORDS(dataBase)\n",
+    "\n",
+    "nb = NaiveBayes(dataBase)\n",
+    "nb.predict(\"Chinese The Chinese In Chinese Tokyo The In The Japan\")\n",
+    "#nb.fit()"
+   ]
+  }
+ ],
+ "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": "3.7.1"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}
diff --git a/Naive_Bayes_text_classification/Plan_Report_NaiveBayes.docx b/Naive_Bayes_text_classification/Plan_Report_NaiveBayes.docx
diff --git a/Naive_Bayes_text_classification/Plan_Report_NaiveBayes.pdf b/Naive_Bayes_text_classification/Plan_Report_NaiveBayes.pdf
diff --git a/Naive_Bayes_text_classification/README.md b/Naive_Bayes_text_classification/README.md
@@ -0,0 +1,7 @@
+In text classification, the goal is to find the best class for the document.
+Multinomial Naive Bayes or multinomial NB model, a probabilistic learning method is a supervised learning method.
+
+Please, use the pdf file for more detailed instructions. For worked example look at 44 slide.
+Use Naive Bayes model with TF/IDF algorithm to solve text classification problem. Be free to choose any task for applying this algorithm.
+The datasets can be taken from broad set of links e.g., kaggle, etc.
+For testing purposes you can use email spam filter dataset from here.
diff --git a/Naive_Bayes_text_classification/Report_NaiveBayes.pdf b/Naive_Bayes_text_classification/Report_NaiveBayes.pdf
diff --git a/Naive_Bayes_text_classification/naivebayes.pdf b/Naive_Bayes_text_classification/naivebayes.pdf