Python_2019-09-03_机器学习——贝叶斯（一）——文本分类

 目录

 思路

bayes.py程序

---

思路

'''
字符串转单词表
词表转词向量0/1
词向量计算概率p(ci)和p(w|ci)
最后计算类别概率：
p(ci|w) = p(w|ci)*p(ci) = log(p(w|ci)) + log(p(ci))
得到最大概率的预测结果
'''

'''
p(ci)的计算
数据+标签：
    [0,1,0,1,0,0,1][1]
    [1,1,0,0,0,0,0][0]
    [0,1,1,1,1,1,1][1]
p(ci)的计算结果
    p(c0)=1/3
    p(c1)=2/3
'''

'''
p(w|ci)的计算
数据+标签：
    [0,1,0,1,0,0,1][1]
    [1,1,0,0,0,0,0][0]
    [0,1,1,1,1,1,1][1]
类别为[1]:
    p1Num1 = p1Num0 + [0,1,0,1,0,0,1] = [1,2,1,2,1,1,2]
    p1Denom1 = p1Denom0 + sum([0,1,0,1,0,0,1]) = 2 + 3
    p1Num2 = p1Num1 + [0,1,1,1,1,1,1] = [1,3,2,3,2,2,3]
    p1Denom2 = p1Denom1 + sum([0,1,1,1,1,1,1]) = 5 + 6 = 11
类别为[0]:
    p0Num1 = p0Num0 + [1,1,0,0,0,0,0] = [2,2,1,1,1,1,1]
    p0Denom1 = p0Denom0 + sum([1,1,0,0,0,0,0]) = 2 + 2 = 4
p(w|ci)的计算结果
    p(w|c0) = log(p0Num1/p0Denom1) = log([2,2,1,1,1,1,1]/04) = [pwc0]
    p(w|c1) = log(p1Num2/p1Denom2) = log([1,3,2,3,2,2,3]/11) = [pwc1]
''' 

'''
测试事件的概率计算p(ci|w) = p(w|ci)*p(ci) = log(p(w|ci)) + log(p(ci))：    
    p(c0|w) = sum([pwc0]*存在类别0的特征向量)+log(p(c0))
    p(c1|w) = sum([pwc1]*存在类别1的特征向量)+log(p(c1))
'''

bayes.py程序

'''
数据样本_第一步：句子+标签
'''
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
'''
数据样本_第二步：句子+标签转单词列表 —— 即【 单词表 】
'''
def createVocabList(dataSet):
    vocabSet = set([])  # create empty set
    for document in dataSet:
        print("dataSet的行document", document)
        print("每句话（每行）set不重复", set(document))
        vocabSet = vocabSet | set(document) # union of the two sets 并集OR 两个set函数定义的集合可以直接计算并、交集，
        print("并集后", vocabSet)
    return list(vocabSet)
'''
数据样本_第三步：单词列表转同等长度的词向量，如果A句子的单词出现则，对应词向量的位置为1，否则为0 —— 即【 单词状态表 】
'''
def setOfWords2Vec(vocabList, inputSet):
    returnVec = [0]*len(vocabList) # 和词列表一样长度
    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

from numpy import *
def trainNB0(trainMatrix, trainCategory):
    numTrainDocs = len(trainMatrix) # 6个数据集
    numWords = len(trainMatrix[0]) # 32个特征词
    '''
    p(ci)的计算
    数据+标签：
        [0,1,0,1,0,0,1][1]
        [1,1,0,0,0,0,0][0]
        [0,1,1,1,1,1,1][1]
    p(ci)的计算结果
        p(c0)=1/3
        p(c1)=2/3
    '''
    pAbusive = sum(trainCategory)/float(numTrainDocs) # 类别为 1 的概率 = sum(类别为1)/总数据集个数
    p0Num = ones(numWords); p1Num = ones(numWords)      #change to ones() 
    p0Denom = 2.0; p1Denom = 2.0                        #change to 2.0
    '''
    p(w|ci)的计算
    数据+标签：
        [0,1,0,1,0,0,1][1]
        [1,1,0,0,0,0,0][0]
        [0,1,1,1,1,1,1][1]
    类别为[1]:
        p1Num1 = p1Num0 + [0,1,0,1,0,0,1] = [1,2,1,2,1,1,2]
        p1Denom1 = p1Denom0 + sum([0,1,0,1,0,0,1]) = 2 + 3
        p1Num2 = p1Num1 + [0,1,1,1,1,1,1] = [1,3,2,3,2,2,3]
        p1Denom2 = p1Denom1 + sum([0,1,1,1,1,1,1]) = 5 + 6 = 11
    类别为[0]:
        p0Num1 = p0Num0 + [1,1,0,0,0,0,0] = [2,2,1,1,1,1,1]
        p0Denom1 = p0Denom0 + sum([1,1,0,0,0,0,0]) = 2 + 2 = 4
    p(w|ci)的计算结果
        p(w|c0) = log(p0Num1/p0Denom1) = log([2,2,1,1,1,1,1]/04) = [pwc0]
        p(w|c1) = log(p1Num2/p1Denom2) = log([1,3,2,3,2,2,3]/11) = [pwc1]
    ''' 
    for i in range(numTrainDocs):
        if trainCategory[i] == 1:   # 类别为 1 的 每个词条进行计算
            p1Num += trainMatrix[i] # 对应位置如果出现，即1，该位置自加。
            p1Denom += sum(trainMatrix[i])  # 对应的总词数也进行统计。
            print(trainMatrix[i])
            print(sum(trainMatrix[i]))
        else:
            p0Num += trainMatrix[i]
            p0Denom += sum(trainMatrix[i])
    p1Vec = log(p1Num/p1Denom)          #change to log()
    p0Vec = log(p0Num/p0Denom)          #change to log()
    return p0Vec, p1Vec, pAbusive

    '''
    测试事件的概率计算p(ci|w) = p(w|ci)*p(ci) = log(p(w|ci)) + log(p(ci))：    
        p(c0|w) = sum([pwc0]*存在类别0的特征向量)+log(p(c0))
        p(c1|w) = sum([pwc1]*存在类别1的特征向量)+log(p(c1))
    '''
def classifyNB(vec2Classify, p0Vec, p1Vec, pClass1):
    p1 = sum(vec2Classify * p1Vec) + log(pClass1)    # p(w|ci)p(ci)在自然对数里+即表示乘法，因为p(w)一样所以就不必计算，只需计算分子
    p0 = sum(vec2Classify * p0Vec) + log(1.0 - pClass1)
    print(p1)
    print(p0)
    if p1 > p0:
        return 1
    else: 
        return 0

if __name__ == "__main__":
    listOPosts, listClasses = loadDataSet()
    print("————————————————————————————————————————————————————", "输入数据", "输入标签")
    print(listOPosts, listClasses)
    
    myVocabList = createVocabList(listOPosts)
    print("————————————————————————————————————————————————————", "转换称词表")
    print(myVocabList)
    
    trainMat = []
    for postinDoc in listOPosts:
        trainMat.append(setOfWords2Vec(myVocabList, postinDoc))
        pass
    print("————————————————————————————————————————————————————", "训练数据和标签")
    for trainMat_rows in trainMat:
        print(trainMat_rows)
        pass
    for class_rows in listClasses:
        print(class_rows)
        pass
    
    p0V, p1V, pAb = trainNB0(array(trainMat),array(listClasses))
    print("————————————————————————————————————————————————————", "每个句子对应的特征向量标号")
    print(p0V)
    print(p1V)
    print(pAb)

    testEntry = ['stupid', 'my']
    thisDoc = array(setOfWords2Vec(myVocabList, testEntry))
    print("————————————————————————————————————————————————————", "测试——每个句子对应的特征向量标号")
    print(thisDoc)    
    print (testEntry,'classified as: ',classifyNB(thisDoc, p0V, p1V, pAb))

    pass