朴素贝叶斯(NaiveBayes)针对小数据集中文文本分类预测

转自相国大人的博客，

http://blog.csdn.net/github_36326955/article/details/54891204

做个笔记

代码按照1 2 3 4的顺序进行即可：

1.py(corpus_segment.py)

#!/usr/bin/env python# -*- coding: UTF-8 -*-"""@version: python2.7.8 @author: XiangguoSun@contact: [e-mail]@file: corpus_segment.py@time: 2017/2/5 15:28@software: PyCharm"""import sysimport osimport jieba# 配置utf-8输出环境reload(sys)sys.setdefaultencoding('utf-8')# 保存至文件def savefile(savepath, content):    with open(savepath, "wb") as fp:        fp.write(content)    '''    上面两行是python2.6以上版本增加的语法，省略了繁琐的文件close和try操作    2.5版本需要from __future__ import with_statement    新手可以参考这个链接来学习http://zhoutall.com/archives/325    '''# 读取文件def readfile(path):    with open(path, "rb") as fp:        content = fp.read()    return contentdef corpus_segment(corpus_path, seg_path):    '''    corpus_path是未分词语料库路径    seg_path是分词后语料库存储路径    '''    catelist = os.listdir(corpus_path)  # 获取corpus_path下的所有子目录    '''    其中子目录的名字就是类别名，例如：    train_corpus/art/21.txt中，'train_corpus/'是corpus_path，'art'是catelist中的一个成员    '''    # 获取每个目录（类别）下所有的文件    for mydir in catelist:        '''        这里mydir就是train_corpus/art/21.txt中的art（即catelist中的一个类别）        '''        class_path = corpus_path + mydir + "/"  # 拼出分类子目录的路径如：train_corpus/art/        seg_dir = seg_path + mydir + "/"  # 拼出分词后存贮的对应目录路径如：train_corpus_seg/art/        if not os.path.exists(seg_dir):  # 是否存在分词目录，如果没有则创建该目录            os.makedirs(seg_dir)        file_list = os.listdir(class_path)  # 获取未分词语料库中某一类别中的所有文本        '''        train_corpus/art/中的        21.txt,        22.txt,        23.txt        ...        file_list=['21.txt','22.txt',...]        '''        for file_path in file_list:  # 遍历类别目录下的所有文件            fullname = class_path + file_path  # 拼出文件名全路径如：train_corpus/art/21.txt            content = readfile(fullname)  # 读取文件内容            '''此时，content里面存贮的是原文本的所有字符，例如多余的空格、空行、回车等等，            接下来，我们需要把这些无关痛痒的字符统统去掉，变成只有标点符号做间隔的紧凑的文本内容            '''            content = content.replace("\r\n", "")  # 删除换行            content = content.replace(" ", "")#删除空行、多余的空格            content_seg = jieba.cut(content)  # 为文件内容分词            savefile(seg_dir + file_path, " ".join(content_seg))  # 将处理后的文件保存到分词后语料目录    print "中文语料分词结束！！！"'''如果你对if __name__=="__main__":这句不懂，可以参考下面的文章http://imoyao.lofter.com/post/3492bc_bd0c4ce简单来说如果其他python文件调用这个文件的函数，或者把这个文件作为模块导入到你的工程中时，那么下面的代码将不会被执行，而如果单独在命令行中运行这个文件，或者在IDE（如pycharm）中运行这个文件时候，下面的代码才会运行。即，这部分代码相当于一个功能测试。如果你还没懂，建议你放弃IT这个行业。'''if __name__=="__main__":    #对训练集进行分词    corpus_path = "/home/appleyuchi/PycharmProjects/MultiNB/csdn_blog/54891204_tenwhy/chinese_text_classification-master/train/"  # 未分词分类语料库路径    seg_path = "/home/appleyuchi/PycharmProjects/MultiNB/csdn_blog/54891204_tenwhy/chinese_text_classification-master/train_corpus_seg/"  # 分词后分类语料库路径,本程序输出结果    corpus_segment(corpus_path,seg_path)    #对测试集进行分词    corpus_path = "/home/appleyuchi/PycharmProjects/MultiNB/csdn_blog/54891204_tenwhy/chinese_text_classification-master/answer/"  # 未分词分类语料库路径    seg_path = "/home/appleyuchi/PycharmProjects/MultiNB/csdn_blog/54891204_tenwhy/chinese_text_classification-master/test_corpus_seg/"  # 分词后分类语料库路径，本程序输出结果    corpus_segment(corpus_path,seg_path)

2.py(corpus2Bunch.py)

#!/usr/bin/env python# -*- coding: UTF-8 -*-"""@version: python2.7.8 @author: XiangguoSun@contact: [e-mail]@file: corpus2Bunch.py@time: 2017/2/7 7:41@software: PyCharm"""import sysreload(sys)sys.setdefaultencoding('utf-8')import os#python内置的包，用于进行文件目录操作，我们将会用到os.listdir函数import cPickle as pickle#导入cPickle包并且取一个别名pickle'''事实上python中还有一个也叫作pickle的包，与这里的名字相同了，无所谓关于cPickle与pickle，请参考博主另一篇博文：python核心模块之pickle和cPickle讲解http://blog.csdn.net/github_36326955/article/details/54882506本文件代码下面会用到cPickle中的函数cPickle.dump'''from sklearn.datasets.base import Bunch#这个您无需做过多了解，您只需要记住以后导入Bunch数据结构就像这样就可以了。#今后的博文会对sklearn做更有针对性的讲解def _readfile(path):    '''读取文件'''    #函数名前面带一个_,是标识私有函数    # 仅仅用于标明而已，不起什么作用，    # 外面想调用还是可以调用，    # 只是增强了程序的可读性    with open(path, "rb") as fp:#with as句法前面的代码已经多次介绍过，今后不再注释        content = fp.read()    return contentdef corpus2Bunch(wordbag_path,seg_path):    catelist = os.listdir(seg_path)# 获取seg_path下的所有子目录，也就是分类信息    #创建一个Bunch实例    bunch = Bunch(target_name=[], label=[], filenames=[], contents=[])    bunch.target_name.extend(catelist)    '''    extend(addlist)是python list中的函数，意思是用新的list（addlist）去扩充    原来的list    '''    # 获取每个目录下所有的文件    for mydir in catelist:        class_path = seg_path + mydir + "/"  # 拼出分类子目录的路径        file_list = os.listdir(class_path)  # 获取class_path下的所有文件        for file_path in file_list:  # 遍历类别目录下文件            fullname = class_path + file_path  # 拼出文件名全路径            bunch.label.append(mydir)            bunch.filenames.append(fullname)            bunch.contents.append(_readfile(fullname))  # 读取文件内容            '''append(element)是python list中的函数，意思是向原来的list中添加element，注意与extend()函数的区别'''    # 将bunch存储到wordbag_path路径中    with open(wordbag_path, "wb") as file_obj:        pickle.dump(bunch, file_obj)    print "构建文本对象结束！！！"if __name__ == "__main__":#这个语句前面的代码已经介绍过，今后不再注释    #对训练集进行Bunch化操作：    wordbag_path = "/home/appleyuchi/PycharmProjects/MultiNB/csdn_blog/54891204_tenwhy/chinese_text_classification-master/train_word_bag/train_set.dat"  # Bunch存储路径，程序输出    seg_path = "/home/appleyuchi/PycharmProjects/MultiNB/csdn_blog/54891204_tenwhy/chinese_text_classification-master/train_corpus_seg/"  # 分词后分类语料库路径，程序输入    corpus2Bunch(wordbag_path, seg_path)    # 对测试集进行Bunch化操作：    wordbag_path = "/home/appleyuchi/PycharmProjects/MultiNB/csdn_blog/54891204_tenwhy/chinese_text_classification-master/test_word_bag/test_set.dat"  # Bunch存储路径，程序输出    seg_path = "/home/appleyuchi/PycharmProjects/MultiNB/csdn_blog/54891204_tenwhy/chinese_text_classification-master/test_corpus_seg/"  # 分词后分类语料库路径，程序输入    corpus2Bunch(wordbag_path, seg_path)

3.py(TFIDF_space.py)

#!/usr/bin/env python# -*- coding: UTF-8 -*-"""@version: python2.7.8 @author: XiangguoSun@contact: [e-mail]@file: TFIDF_space.py@time: 2017/2/8 11:39@software: PyCharm"""import sysreload(sys)sys.setdefaultencoding('utf-8')from sklearn.datasets.base import Bunchimport cPickle as picklefrom sklearn.feature_extraction.text import TfidfVectorizerdef _readfile(path):    with open(path, "rb") as fp:        content = fp.read()    return contentdef _readbunchobj(path):    with open(path, "rb") as file_obj:        bunch = pickle.load(file_obj)    return bunchdef _writebunchobj(path, bunchobj):    with open(path, "wb") as file_obj:        pickle.dump(bunchobj, file_obj)def vector_space(stopword_path,bunch_path,space_path,train_tfidf_path=None):    stpwrdlst = _readfile(stopword_path).splitlines()    bunch = _readbunchobj(bunch_path)    tfidfspace = Bunch(target_name=bunch.target_name, label=bunch.label, filenames=bunch.filenames, tdm=[], vocabulary={})    if train_tfidf_path is not None:        trainbunch = _readbunchobj(train_tfidf_path)        tfidfspace.vocabulary = trainbunch.vocabulary        vectorizer = TfidfVectorizer(stop_words=stpwrdlst, sublinear_tf=True, max_df=0.5,vocabulary=trainbunch.vocabulary)        tfidfspace.tdm = vectorizer.fit_transform(bunch.contents)    else:        vectorizer = TfidfVectorizer(stop_words=stpwrdlst, sublinear_tf=True, max_df=0.5)        tfidfspace.tdm = vectorizer.fit_transform(bunch.contents)        tfidfspace.vocabulary = vectorizer.vocabulary_    _writebunchobj(space_path, tfidfspace)    print "tf-idf词向量空间实例创建成功！！！"if __name__ == '__main__':    # stopword_path = "/home/appleyuchi/PycharmProjects/MultiNB/csdn_blog/54891204/chinese_text_classification-master/train_word_bag/hlt_stop_words.txt"#输入的文件    # bunch_path = "train_word_bag/train_set.dat"#输入的文件    # space_path = "train_word_bag/tfdifspace.dat"#输出的文件    # vector_space(stopword_path,bunch_path,space_path)    #    # bunch_path = "test_word_bag/test_set.dat"#输入的文件    # space_path = "test_word_bag/testspace.dat"    # train_tfidf_path="train_word_bag/tfdifspace.dat"    # vector_space(stopword_path,bunch_path,space_path,train_tfidf_path)    stopword_path = "/home/appleyuchi/PycharmProjects/MultiNB/csdn_blog/54891204_tenwhy/chinese_text_classification-master/train_word_bag/hlt_stop_words.txt"#输入的文件    train_bunch_path = "/home/appleyuchi/PycharmProjects/MultiNB/csdn_blog/54891204_tenwhy/chinese_text_classification-master/train_word_bag/train_set.dat"#输入的文件    space_path = "/home/appleyuchi/PycharmProjects/MultiNB/csdn_blog/54891204_tenwhy/chinese_text_classification-master/train_word_bag/tfidfspace.dat"#输出的文件    vector_space(stopword_path,train_bunch_path,space_path)    train_tfidf_path = "/home/appleyuchi/PycharmProjects/MultiNB/csdn_blog/54891204_tenwhy/chinese_text_classification-master/train_word_bag/tfidfspace.dat"  # 输入的文件，由上面生成    test_bunch_path = "/home/appleyuchi/PycharmProjects/MultiNB/csdn_blog/54891204_tenwhy/chinese_text_classification-master/test_word_bag/test_set.dat"#输入的文件    test_space_path = "/home/appleyuchi/PycharmProjects/MultiNB/csdn_blog/54891204_tenwhy/chinese_text_classification-master/test_word_bag/testspace.dat"#输出的文件    vector_space(stopword_path,test_bunch_path,test_space_path,train_tfidf_path)

4.py(NBayes_Predict.py)

#!/usr/bin/env python# -*- coding: UTF-8 -*-"""@version: python2.7.8 @author: XiangguoSun@contact: [e-mail]@file: NBayes_Predict.py@time: 2017/2/8 12:21@software: PyCharm"""import sysreload(sys)sys.setdefaultencoding('utf-8')import cPickle as picklefrom sklearn.naive_bayes import MultinomialNB  # 导入多项式贝叶斯算法# 读取bunch对象def _readbunchobj(path):    with open(path, "rb") as file_obj:        bunch = pickle.load(file_obj)    return bunch# 导入训练集trainpath = "/home/appleyuchi/PycharmProjects/MultiNB/csdn_blog/54891204_tenwhy/chinese_text_classification-master/train_word_bag/tfidfspace.dat"train_set = _readbunchobj(trainpath)# 导入测试集testpath = "/home/appleyuchi/PycharmProjects/MultiNB/csdn_blog/54891204_tenwhy/chinese_text_classification-master/test_word_bag/testspace.dat"test_set = _readbunchobj(testpath)# 训练分类器：输入词袋向量和分类标签，alpha:0.001 alpha越小，迭代次数越多，精度越高clf = MultinomialNB(alpha=0.01).fit(train_set.tdm, train_set.label)# 预测分类结果predicted = clf.predict(test_set.tdm)for flabel,file_name,expct_cate in zip(test_set.label,test_set.filenames,predicted):    if flabel != expct_cate:        print file_name,": 实际类别:",flabel," -->预测类别:",expct_cateprint "预测完毕!!!"# 计算分类精度：from sklearn import metricsdef metrics_result(actual, predict):    print '精度:{0:.3f}'.format(metrics.precision_score(actual, predict,average='weighted'))    print '召回:{0:0.3f}'.format(metrics.recall_score(actual, predict,average='weighted'))    print 'f1-score:{0:.3f}'.format(metrics.f1_score(actual, predict,average='weighted'))metrics_result(test_set.label, predicted)

大概说下用法：

一、上面四个代码依次运行即可

二、要注意数据的存放方式要和转载的博客中一样，文件夹的名字就是类别名字，代码会进行自动识别。

三、每次跑完一遍流程，跑下一次程序前，train_corpus_seg和test_corpus_seg两个文件夹要全部删除，不然上次残留的结果会影响这次的预测。

同样地，如果更换中文数据集，这两个文件夹也要删除，总之，运行以上代码的第一步骤就是检查这两个文件夹下面是不是空的。（当然如果是第一次运行以上四个代码，没有生成这两个文件夹，自然是不用检查的）

另外，他这篇博客的优点是，可以针对小数据集（数据条数不到1000，十折交叉验证），预测概率可以达到60%~70%

程序之间的输入输出关系图