#!/usr/bin/python#-*- coding: utf-8 -*-##########################################Bayes : 用来描述两个条件概率之间的关系

#参数: inX: vector to compare to existing dataset (1xN)#dataSet: size m data set of known vectors (NxM)#labels: data set labels (1xM vector)#公式：P(A|B)=P(B|A)P(A)/P(B)#输出: 出错率#########################################

importnumpy as npyimportosimporttime#P(B|A)=P(A|B)*P(A)/P(B)

#数据集目录

dataSetDir ='E:/digits/'

classBayes:def __init__(self):

self.length=-1self.labelrate=dict()

self.vectorrate=dict()deffit(self,dataset:list,labels:list):print("训练开始")if len(dataset)!=len(labels):raise ValueError("输入测试数组和类别数组长度不一致")

self.length=len(dataset[0])#训练数据特征值的长度

labelsnum=len(labels) #类别的数量

norlabels=set(labels) #不重复类别的数量

for item innorlabels:

self.labelrate[item]=labels.count(item)/labelsnum #求当前类别占总类别的比例

for vector,label inzip(dataset,labels):if label not inself.vectorrate:

self.vectorrate[label]=[]

self.vectorrate[label].append(vector)print("训练结束")returnselfdefbtest(self,testdata,labelset):if self.length==-1:raise ValueError("未开始训练，先训练")#计算testdata分别为各个类别的概率

lbDict=dict()for thislb inlabelset:

p= 1alllabel=self.labelrate[thislb]

allvector=self.vectorrate[thislb]

vnum=len(allvector)

allvector=npy.array(allvector).Tfor index inrange(0,len(testdata)):

vector=list(allvector[index])

p*=vector.count(testdata[index])/vnum

lbDict[thislb]=p *alllabel

thislbabel=sorted(lbDict,key=lambda x:lbDict[x],reverse=True)[0]returnthislbabel#加载数据

defdatatoarray(fname):

arr=[]

fh=open(fname)for i in range(0,32):

thisline=fh.readline()for j in range(0 , 32):

arr.append(int(thisline[j]))returnarr#建立一个函数取出labels

defseplabel(fname):

filestr=fname.split(".")[0]

label=int(filestr.split("_")[0])returnlabel#建立训练数据

deftraindata():

labels=[]

trainfile=os.listdir(dataSetDir+"trainingDigits") #加载测试数据

num=len(trainfile)

trainarr=npy.zeros((num,1024))for i inrange(num):

thisfname=trainfile[i]

thislabel=seplabel(thisfname)

labels.append(thislabel)

trainarr[i,]=datatoarray(dataSetDir+"trainingDigits/"+thisfname)returntrainarr,labels#贝叶斯算法手写识别主流程

bys=Bayes()

start=time.time()## step 1: 训练数据集

train_data,labels=traindata()

train_data=list(train_data)

bys.fit(train_data,labels)## step 2:测试数据集

thisdata=datatoarray(dataSetDir+"testDigits/8_90.txt")

labelsall=[0,1,2,3,4,5,6,7,8,9]## 识别单个手写体数字#test=bys.btest(thisdata,labelsall)#print(test)

## 识别多个手写体数字(批量处理),并输出结果

testfile=os.listdir(dataSetDir+"testDigits")

num=len(testfile)

x=0for i inrange(num):

thisfilename=testfile[i]

thislabel=seplabel(thisfilename)

thisdataarr=datatoarray(dataSetDir+"testDigits/"+thisfilename)

label=bys.btest(thisdataarr,labelsall)print("测试数字是："+str(thislabel)+"识别出来的数字是："+str(label))if label!=thislabel:

x+=1

print("识别出错")print(x)print("出错率："+str(x/num))

end=time.time()

running_time= end-startprint('程序运行总耗时： %.5f sec' %running_time)