【python】python机器学习实战3：朴素贝叶斯分类器-北京尧图网络

1.朴素贝叶斯简介第一个问题为什么叫“朴素”是因为在整个形式化过程只做最原始、最简单的假设。朴素贝叶斯的优点在数据较少的情况下仍然有效可以处理多类别问题。缺点是对于输入数据的准备方式较为敏感适用的数据类型标称型数据。网上基本的介绍贝叶斯的理论已经很多了这里我就不再重复赘述了。LZ分享一下原始代码和数据集有兴趣的小伙伴可以自行下载本代码在python2.7版本下运行是没有问题的如果是python3.X版本的可能有些地方会报错这个代码修改请小伙伴自行解决了。链接: https://pan.baidu.com/s/1c228QTM 密码: 35yf2.代码实现#coding:utf-8#导入numpy依赖库fromnumpyimport*#词表到向量的转换函数首先要先构建数据集及其对应的标签defloadDataSet():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 notreturnpostingList,classVecdefcreateVocabList(dataSet):vocabSetset([])#create empty setfordocumentindataSet:#操作符|用于两个集合的并集vocabSetvocabSet|set(document)#union of the two setsreturnlist(vocabSet)#转换成词向量长度和输入的单词列表相同如果在文本中出现那个单词则该单词出现的位置值设为1defsetOfWords2Vec(vocabList,inputSet):returnVec[0]*len(vocabList)forwordininputSet:ifwordinvocabList:returnVec[vocabList.index(word)]1else:printthe word: %s is not in my Vocabulary!%wordreturnreturnVec#朴素贝叶斯分类器训练函数deftrainNB0(trainMatrix,trainCategory):numTrainDocslen(trainMatrix)numWordslen(trainMatrix[0])pAbusivesum(trainCategory)/float(numTrainDocs)p0Numones(numWords);p1Numones(numWords)#change to ones()p0Denom2.0;p1Denom2.0#change to 2.0foriinrange(numTrainDocs):iftrainCategory[i]1:p1NumtrainMatrix[i]p1Denomsum(trainMatrix[i])else:p0NumtrainMatrix[i]p0Denomsum(trainMatrix[i])p1Vectlog(p1Num/p1Denom)#change to log()p0Vectlog(p0Num/p0Denom)#change to log()returnp0Vect,p1Vect,pAbusive#朴素贝叶斯分类函数defclassifyNB(vec2Classify,p0Vec,p1Vec,pClass1):p1sum(vec2Classify*p1Vec)log(pClass1)#element-wise multp0sum(vec2Classify*p0Vec)log(1.0-pClass1)ifp1p0:return1else:return0#文档的词袋模型每当遇到一个单词是它会增加词向量中的对应值而不将对应值设为1设为1的是词集模型defbagOfWords2VecMN(vocabList,inputSet):returnVec[0]*len(vocabList)forwordininputSet:ifwordinvocabList:returnVec[vocabList.index(word)]1returnreturnVec#便利函数deftestingNB():listOPosts,listClassesloadDataSet()myVocabListcreateVocabList(listOPosts)trainMat[]forpostinDocinlistOPosts:trainMat.append(setOfWords2Vec(myVocabList,postinDoc))p0V,p1V,pAbtrainNB0(array(trainMat),array(listClasses))testEntry[love,my,dalmation]thisDocarray(setOfWords2Vec(myVocabList,testEntry))printtestEntry,classified as: ,classifyNB(thisDoc,p0V,p1V,pAb)testEntry[stupid,garbage]thisDocarray(setOfWords2Vec(myVocabList,testEntry))printtestEntry,classified as: ,classifyNB(thisDoc,p0V,p1V,pAb)#切分文本#文件解析完整的垃圾邮件测试函数deftextParse(bigString):#input is big string, #output is word listimportre listOfTokensre.split(r\W*,bigString)return[tok.lower()fortokinlistOfTokensiflen(tok)2]#对贝叶斯垃圾邮件分类器进行自动化处理defspamTest():docList[];classList[];fullText[]foriinrange(1,26):wordListtextParse(open(email/spam/%d.txt%i).read())docList.append(wordList)fullText.extend(wordList)classList.append(1)wordListtextParse(open(email/ham/%d.txt%i).read())docList.append(wordList)fullText.extend(wordList)classList.append(0)vocabListcreateVocabList(docList)#create vocabularytrainingSetrange(50);testSet[]#create test setforiinrange(10):randIndexint(random.uniform(0,len(trainingSet)))testSet.append(trainingSet[randIndex])del(trainingSet[randIndex])trainMat[];trainClasses[]fordocIndexintrainingSet:#train the classifier (get probs) trainNB0trainMat.append(bagOfWords2VecMN(vocabList,docList[docIndex]))trainClasses.append(classList[docIndex])p0V,p1V,pSpamtrainNB0(array(trainMat),array(trainClasses))errorCount0fordocIndexintestSet:#classify the remaining itemswordVectorbagOfWords2VecMN(vocabList,docList[docIndex])ifclassifyNB(array(wordVector),p0V,p1V,pSpam)!classList[docIndex]:errorCount1printclassification error,docList[docIndex]printthe error rate is: ,float(errorCount)/len(testSet)#return vocabList,fullText#RSS源分类器及高频词去除函数#遍历词汇表中每个词并统计它在文本中出现的次数排序后返回前30个单词defcalcMostFreq(vocabList,fullText):importoperator freqDict{}fortokeninvocabList:freqDict[token]fullText.count(token)sortedFreqsorted(freqDict.iteritems(),keyoperator.itemgetter(1),reverseTrue)returnsortedFreq[:30]deflocalWords(feed1,feed0):importfeedparser docList[];classList[];fullText[]minLenmin(len(feed1[entries]),len(feed0[entries]))foriinrange(minLen):wordListtextParse(feed1[entries][i][summary])docList.append(wordList)fullText.extend(wordList)classList.append(1)#NY is class 1wordListtextParse(feed0[entries][i][summary])docList.append(wordList)fullText.extend(wordList)classList.append(0)vocabListcreateVocabList(docList)#create vocabularytop30WordscalcMostFreq(vocabList,fullText)#remove top 30 wordsforpairWintop30Words:ifpairW[0]invocabList:vocabList.remove(pairW[0])trainingSetrange(2*minLen);testSet[]#create test setforiinrange(20):randIndexint(random.uniform(0,len(trainingSet)))testSet.append(trainingSet[randIndex])del(trainingSet[randIndex])trainMat[];trainClasses[]fordocIndexintrainingSet:#train the classifier (get probs) trainNB0trainMat.append(bagOfWords2VecMN(vocabList,docList[docIndex]))trainClasses.append(classList[docIndex])p0V,p1V,pSpamtrainNB0(array(trainMat),array(trainClasses))errorCount0fordocIndexintestSet:#classify the remaining itemswordVectorbagOfWords2VecMN(vocabList,docList[docIndex])ifclassifyNB(array(wordVector),p0V,p1V,pSpam)!classList[docIndex]:errorCount1printthe error rate is: ,float(errorCount)/len(testSet)returnvocabList,p0V,p1V#最具表征性词汇显示函数defgetTopWords(ny,sf):importoperator vocabList,p0V,p1VlocalWords(ny,sf)topNY[];topSF[]foriinrange(len(p0V)):ifp0V[i]-6.0:topSF.append((vocabList[i],p0V[i]))ifp1V[i]-6.0:topNY.append((vocabList[i],p1V[i]))sortedSFsorted(topSF,keylambdapair:pair[1],reverseTrue)printSF**SF**SF**SF**SF**SF**SF**SF**SF**SF**SF**SF**SF**SF**SF**SF**foriteminsortedSF:printitem[0]sortedNYsorted(topNY,keylambdapair:pair[1],reverseTrue)printNY**NY**NY**NY**NY**NY**NY**NY**NY**NY**NY**NY**NY**NY**NY**NY**foriteminsortedNY:printitem[0]哈哈完美啦O(∩_∩)O

【python】python机器学习实战3：朴素贝叶斯分类器

相关推荐

【数据处理】如何处理数据中的缺失值？？

【android】三方库的Android移植

【android】Application.mk的文件梳理

更多资讯

AI岗位暴涨12倍！小白程序员必备：收藏这份2026年高薪跳槽指南

具身智能协同演化动力学（120）：TVA-World 协同架构虚实域适应机理研究

服务器买了一台又一台？虚拟化到云计算的演进一次讲清

2026年大模型AI校招薪资狂飙！顶尖年薪高达154万，非AI专业也能入行！

企业小白程序员必看：AI 模型如何真正“懂业务”实现落地？

OpenAI 挨个通知了 100 多家组织：给 Agent 补一张出站责任清单

今日更新

Verilog实现FPGA密码锁：矩阵键盘与状态机设计实战

数据服务监控运维实战：从指标体系到告警治理与故障排查

FreeRTOS任务通知实战详解：内核原理、API与避坑指南

本周精选

金融AI落地指南：从调研报告到技术选型与工程实践

大模型上下文成本与注意力优化技术深度解析——为何O(n²)平方复杂度无法被常规优化消除｜TaoToken 统一 Key 通道实测

基于GMM的语音识别系统原理与MATLAB实现详解

本月热门

我发现了一个新思路：用 Remotion + Claude Code 像写代码一样自动化生成短视频

Windows下 Codex 中 Chrome 和 Computer Use 插件不可用问题排查及解决参考方式：TaoToken 统一 Key 配置与验证

2026 大模型集体涨价：用 Python 做企业 Token 成本测算与选型避坑（附配置）

读完文章，想聊聊您的网站？