使用朴素贝叶斯进行文本分类

代码如下：

import numpy as np
def loadDataSet():
    postingList = [['my', 'dog', 'has', 'flea', 'problems', 'help', 'please'],  # 切分的词条
                   ['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]
    return postingList, classVec


# 1.创建词汇表
def creatVocabList(dataSet):
    lists = set([])
    for document in dataSet:
        lists = lists | set(document)
    return list(lists)


# 2.利用词汇表创建文本向量
def setOfWords2Vec(vocalist, input):
    wordlist = [0] * len(vocalist)
    for word in input:
        wordlist[vocalist.index(word)] = 1
    return wordlist


#3.朴素贝叶斯分类器训练函数
#计算公式：p（侮辱类|单词）=p（侮辱类）*p（单词|侮辱类）/p（单词）
#但是因为比较大小，因此分母的单词p不需要去计算
def trainNB0(trainMatrix,trainCategory):
    numTrainDocs=len(trainMatrix)#文档数目
    numWords=len(trainMatrix[0])#文档中词的数量
    pAbusive = sum(trainCategory) / float(numTrainDocs)  # 文档属于侮辱类的概率   p(侮辱类)
    p0Num = np.ones(numWords)
    p1Num = np.ones(numWords)  # 计算侮辱类的情况下每个单词出现的次数或者非侮辱类的，保存数组
    p0Denom = 1.0
    p1Denom = 1.0  # 分母初始化为2，运用优化，拉普拉斯平滑
    for i in range(numTrainDocs):#遍历样本
        if trainCategory[i]==1:
            p1Num += trainMatrix[i]
            p1Denom += sum(trainMatrix[i])
        else:  # 统计属于非侮辱类的条件概率所需的数据，即P(w0|0),P(w1|0),P(w2|0)···
            p0Num += trainMatrix[i]#向量相加，判断每个单词 在每个类别中出现多少次
            p0Denom += sum(trainMatrix[i])#统计单词个数
        p1Vect = np.log(p1Num / p1Denom)
        p0Vect = np.log(p0Num / p0Denom)
        return p0Vect, p1Vect, pAbusive

def classifyNB(vec2Classify, p0Vec, p1Vec, pClass1):
    p1=sum(vec2Classify*p1Vec)+np.log(pClass1)
    p0=sum(vec2Classify*p0Vec)+np.log(1.0-pClass1)
    if p1 > p0:
        return 1
    else:
        return 0

lists, classes = loadDataSet()
alphabet = creatVocabList(lists)
trainMat=[]
for i in lists:
    trainMat.append(setOfWords2Vec(alphabet ,i))
p0v,p1v,pAb=trainNB0(trainMat,classes)
test=['love','my','dalmation']
thisdoc=setOfWords2Vec(alphabet,test)
if classifyNB(thisdoc,p0v,p1v,pAb):
    print(test,'属于侮辱类')										#执行分类并打印分类结果
else:
    print(test,'属于非侮辱类')
test2=['stupid','garbage']
thisdoc2=setOfWords2Vec(alphabet,test2)
if classifyNB(thisdoc2,p0v,p1v,pAb):
    print(test2, '属于侮辱类')  # 执行分类并打印分类结果
else:
    print(test2, '属于非侮辱类')
test3=['mr']
thisdoc3=setOfWords2Vec(alphabet,test3)
if classifyNB(thisdoc3,p0v,p1v,pAb):
    print(test3, '属于侮辱类')  # 执行分类并打印分类结果
else:
    print(test3, '属于非侮辱类')