Python 学习笔记（Machine Learning In Action）K-近邻算法识别手写数字

 1 from numpy import *
 2 import matplotlib
 3 from os import listdir
 4 import kNN
 5 def img2vector(filename):
 6     returnVect = zeros((1,1024))
 7     fr = open(filename)
 8     for i in range(32):
 9         lineStr = fr.readline()
10         for j in range(32):
11             returnVect[0,32*i+j] = int(lineStr[j])
12     return returnVect
13 def handwritingClassTest():
14     hwLabels = []
15     trainingFileList = listdir(‘trainingDigits‘)           #load the training set
16     m = len(trainingFileList)
17     trainingMat = zeros((m,1024))
18     for i in range(m):
19         fileNameStr = trainingFileList[i]
20         fileStr = fileNameStr.split(‘.‘)[0]       #take off .txt     0_0.txt 将整个文件名以 . 分开 取第一部分也就是0_0
21         classNumStr = int(fileStr.split(‘_‘)[0])  #将 0_0 以 _ 分开 取第一部分也就是0
22         hwLabels.append(classNumStr)
23         trainingMat[i,:] = img2vector(‘trainingDigits/%s‘ % fileNameStr)
24     testFileList = listdir(‘testDigits‘)        #iterate through the test set
25     errorCount = 0.0
26     mTest = len(testFileList)
27     for i in range(mTest):
28         fileNameStr = testFileList[i]
29         fileStr = fileNameStr.split(‘.‘)[0]     #take off .txt
30         classNumStr = int(fileStr.split(‘_‘)[0])
31         vectorUnderTest = img2vector(‘testDigits/%s‘ % fileNameStr)
32         classifierResult = kNN.classify0(vectorUnderTest, trainingMat, hwLabels, 3)
33         print "the classifier came back with: %d, the real answer is: %d" % (classifierResult, classNumStr)
34         if (classifierResult != classNumStr): errorCount += 1.0
35     print "\nthe total number of errors is: %d" % errorCount
36     print "\nthe total error rate is: %f" % (errorCount/float(mTest))
37 handwritingClassTest()