Python Bayes

from numpy import *
    
def loadDataSet():
    postingList=[['my', 'dog', 'has', 'flea', 'problems', 'help', 'please'],
                 ['maybe', 'not', 'take', 'him', 'to', 'dog', 'park', 'stupid'],
                 ['my', 'dalmation', 'is', 'so', 'cute', 'I', 'love', 'him'],
                 ['stop', 'posting', 'stupid', 'worthless', 'garbage'],
                 ['mr', 'licks', 'ate', 'my', 'steak', 'how', 'to', 'stop', 'him'],
                 ['quit', 'buying', 'worthless', 'dog', 'food', 'stupid']]
    classVec = [0,1,0,1,0,1]    #1 is abusive, 0 not
    return postingList,classVec
         
         
                  
def createVocabList(dataSet):
    vocabSet = set([])  #create empty set
    for document in dataSet:
        vocabSet = vocabSet | set(document) #union of the two sets
    return list(vocabSet)
 
 
def setOfWords2Vec(vocabList, inputSet):
    returnVec = [0]*len(vocabList)
    for word in inputSet:
        if word in vocabList:
            returnVec[vocabList.index(word)] = 1
        else: print "the word: %s is not in my Vocabulary!" % word
    return returnVec
 
 
def trainNB0(trainMatrix,trainCategory):
    #numTrainDocs训练集总条数
    numTrainDocs = len(trainMatrix)
    #每条训练数据总单词数
    numWords = len(trainMatrix[0])
    #侮辱类的概率(侮辱类占总训练数据的比例)
    pAbusive = sum(trainCategory)/float(numTrainDocs)
    
    #拉普拉斯平滑，所有单词一开始全部置为1，防止概率为0
    #正常类向量置为1
    p0Num = ones(numWords);
    #侮辱类向量置为1    
    p1Num = ones(numWords) 
    #分母置为2    
    p0Denom = 2.0; 
    p1Denom = 2.0               
    #遍历训练集数据    
    for i in range(numTrainDocs):
        #该条训练数据为侮辱类
        if trainCategory[i] == 1:
            #侮辱类所含单词，次数加1
            p1Num += trainMatrix[i]
            #p1Denom侮辱类总词数
            p1Denom += sum(trainMatrix[i])
            
        #该条训练数据为正常类
        else:
            #正常类所含单词，次数增加
            p0Num += trainMatrix[i]
            #p0Denom正常类总词数
            p0Denom += sum(trainMatrix[i])   
    #数据取log，即单个单词的p(x1|c1)取log，防止下溢出        
    p1Vect = log(p1Num/p1Denom)         
    p0Vect = log(p0Num/p0Denom) 
    #返回正常类、侮辱类的类条件概率向量，p0Vect,p1Vect
    #每一项为拉普拉斯平滑处理后词的类条件概率取log以2底
    #返回侮辱类的概率pAbusive
    return p0Vect,p1Vect,pAbusive
 
def classifyNB(vec2Classify, p0Vec, p1Vec, pClass1):
    #在对数空间中进行计算，属于哪一类的概率比较大就判为哪一类
    p1 = sum(vec2Classify * p1Vec) + log(pClass1)    
    p0 = sum(vec2Classify * p0Vec) + log(1.0 - pClass1)   
    if p1 > p0:
        return 1
    else: 
        return 0
 
def testingNB():
    #获得训练数据，类别标签
    listOPosts,listClasses = loadDataSet()
    #创建词典
    myVocabList = createVocabList(listOPosts)
    #构建矩阵，存放训练数据
    trainMat=[]
    
    #遍历原始数据，转换为词向量，构成数据训练矩阵
    for postinDoc in listOPosts:
        #数据转换后存入数据训练矩阵trainMat中
        trainMat.append(setOfWords2Vec(myVocabList, postinDoc))
    #训练分类器
    p0V,p1V,pAb = trainNB0(array(trainMat),array(listClasses))
    
    #测试
    testEntry = ['love', 'my', 'dalmation']
    #测试数据转为词向量
    thisDoc = array(setOfWords2Vec(myVocabList, testEntry))
    #输出分类结果
    print testEntry,'classified as: ',classifyNB(thisDoc,p0V,p1V,pAb)
    
    #测试
    testEntry = ['stupid', 'garbage']
    #测试数据转为词向量
    thisDoc = array(setOfWords2Vec(myVocabList, testEntry))
    #输出分类结果
    print testEntry,'classified as: ',classifyNB(thisDoc,p0V,p1V,pAb)
 
 
#文档词袋模型，词向量 值为该词在本篇文章中出现的次数
def bagOfWords2VecMN(vocabList, inputSet):
    returnVec = [0]*len(vocabList)
    for word in inputSet:
        if word in vocabList:
            returnVec[vocabList.index(word)] += 1
    return returnVec
 

#准备数据，按空格切分出词 
#单词长度小于或等于2的全部丢弃
def textParse(bigString):    
    import re
    listOfTokens = re.split(r'\W*', bigString)
    #tok.lower() 将整个词转换为小写
    return [tok.lower() for tok in listOfTokens if len(tok) > 2] 
     
def spamTest():
    #文章按篇存放
    docList=[]; 
    #存放文章类别
    classList = [];
    #存放所有文章内容    
    fullText =[]
    for i in range(1,26):
        #读取垃圾邮件
        wordList = textParse(open('D:/email/spam/%d.txt' % i).read())
        #docList按篇存放文章
        docList.append(wordList)
        #fullText邮件内容存放到一起
        fullText.extend(wordList)
        #垃圾邮件类别标记为1
        classList.append(1)
        
        #读取正常邮件
        wordList = textParse(open('D:/email/ham/%d.txt' % i).read())
        docList.append(wordList)
        fullText.extend(wordList)
        #正常邮件类别标记为0
        classList.append(0)
        
    #创建词典    
    vocabList = createVocabList(docList)
    #训练集共50篇文章
    trainingSet = range(50);
    #创建测试集
    testSet=[]
    #随机选取10篇文章为测试集，测试集中文章从训练集中删除    
    for i in range(10):
        #0-50间产生一个随机数
        randIndex = int(random.uniform(0,len(trainingSet)))
        #从训练集中找到对应文章，加入测试集中
        testSet.append(trainingSet[randIndex])
        #删除对应文章
        del(trainingSet[randIndex])  
        
    #准备数据，用于训练分类器    
    trainMat=[]; #训练数据
    trainClasses = [] #类别标签
    
    #遍历训练集中文章数据
    for docIndex in trainingSet:
        #每篇文章转为词袋向量模型，存入trainMat数据矩阵中
        trainMat.append(bagOfWords2VecMN(vocabList, docList[docIndex]))
        #trainClasses存放每篇文章的类别
        trainClasses.append(classList[docIndex])
    #训练分类器
    p0V,p1V,pSpam = trainNB0(array(trainMat),array(trainClasses))
    
    #errorCount记录测试数据出错次数
    errorCount = 0
    #遍历测试数据集，每条数据相当于一条文本
    for docIndex in testSet:
        #文本转换为词向量模型    
        wordVector = bagOfWords2VecMN(vocabList, docList[docIndex])
        #模型给出的分类结果与本身类别不一致时，说明模型出错，errorCount数加1
        if classifyNB(array(wordVector),p0V,p1V,pSpam) != classList[docIndex]:
            errorCount += 1
            #输出出错的文章
            print "classification error",docList[docIndex]
    
    #输出错误率，即出错次数/总测试次数
    print 'the error rate is: ',float(errorCount)/len(testSet)
    
    
    #return vocabList,fullText
     
if __name__ == "__main__":
    #获取数据
    listOPosts,listClasses = loadDataSet()  
    #构建词典
    myVocabList = createVocabList(listOPosts)
    trainMat = []
    for postinDoc in listOPosts:
        #构建训练矩阵
        trainMat.append(setOfWords2Vec(myVocabList, postinDoc))
    p0V,p1V,pAb = trainNB0(trainMat, listClasses)
    #网站论坛内容分类    
    testingNB()
    
    #垃圾邮件分类
    spamTest()