KNN算法的感受 1

本来预计的打算是一天一个十大挖掘算法，然而由于同时要兼顾数据结构面试的事情，所以 很难办到，但至少在回家前要把数据挖掘十大算法看完，过个好年，在course上学习老吴的课程还是帮了我很大的忙，虽然浪费了时间，但是也无形中帮助我 很多，所以说还是很值得的，今天就总结KNN算法的一部分，这部分老吴的课程中没有太多涉及到，所以我又重新关注了一下，下面是我的总结，希望能对大家有 所帮组。

介绍环镜：python2.7  IDLE  Pycharm5.0.3

操作系统：windows

第一步：因为没有numpy，所以要安装numpy，详情见另一篇安装numpy的博客，这里不再多说.

第二步：贴代码：

 1     from numpy import *
 2     import operator
 3     from os import listdir
 5   def classify0(inX, dataSet, labels, k):
 6         dataSetSize = dataSet.shape[0]
 7         diffMat = tile(inX, (dataSetSize,1)) - dataSet
 8         sqDiffMat = diffMat**2
 9         sqDistances = sqDiffMat.sum(axis=1)
10         distances = sqDistances**0.5
11         sortedDistIndicies = distances.argsort()
12         classCount={}
13         for i in range(k):
14             voteIlabel = labels[sortedDistIndicies[i]]
15             classCount[voteIlabel] = classCount.get(voteIlabel,0) + 1
16         sortedClassCount = sorted(classCount.iteritems(), key=operator.itemgetter(1), reverse=True)
17         return sortedClassCount[0][0]
18
19   def createDataSet():
20         group = array([[1.0,1.1],[1.0,1.0],[0,0],[0,0.1]])
21         labels = [‘A‘,‘A‘,‘B‘,‘B‘]
22         return group, labels
23
24    def file2matrix(filename):
25         fr = open(filename)
26         numberOfLines = len(fr.readlines())         #get the number of lines in the file
27         returnMat = zeros((numberOfLines,3))        #prepare matrix to return
28         classLabelVector = []                       #prepare labels return
29         fr = open(filename)
30         index = 0
31         for line in fr.readlines():
32             line = line.strip()
33             listFromLine = line.split(‘\t‘)
34             returnMat[index,:] = listFromLine[0:3]
35             classLabelVector.append(int(listFromLine[-1]))
36             index += 1
37         return returnMat,classLabelVector
38
39     def autoNorm(dataSet):
40         minVals = dataSet.min(0)
41         maxVals = dataSet.max(0)
42         ranges = maxVals - minVals
43         normDataSet = zeros(shape(dataSet))
44         m = dataSet.shape[0]
45         normDataSet = dataSet - tile(minVals, (m,1))
46         normDataSet = normDataSet/tile(ranges, (m,1))   #element wise divide
47         return normDataSet, ranges, minVals
48
49     def datingClassTest():
50         hoRatio = 0.50      #hold out 10%
51         datingDataMat,datingLabels = file2matrix(‘datingTestSet2.txt‘)       #load data setfrom file
52         normMat, ranges, minVals = autoNorm(datingDataMat)
53         m = normMat.shape[0]
54         numTestVecs = int(m*hoRatio)
55         errorCount = 0.0
56         for i in range(numTestVecs):
57             classifierResult = classify0(normMat[i,:],normMat[numTestVecs:m,:],datingLabels[numTestVecs:m],3)
58             print "the classifier came back with: %d, the real answer is: %d" % (classifierResult, datingLabels[i])
59             if (classifierResult != datingLabels[i]): errorCount += 1.0
60         print "the total error rate is: %f" % (errorCount/float(numTestVecs))
61         print errorCount
62
63     def img2vector(filename):
64         returnVect = zeros((1,1024))
65         fr = open(filename)
66         for i in range(32):
67             lineStr = fr.readline()
68             for j in range(32):
69                 returnVect[0,32*i+j] = int(lineStr[j])
70         return returnVect
71
72     def handwritingClassTest():
73         hwLabels = []
74         trainingFileList = listdir(‘trainingDigits‘)           #load the training set
75         m = len(trainingFileList)
76         trainingMat = zeros((m,1024))
77         for i in range(m):
78             fileNameStr = trainingFileList[i]
79             fileStr = fileNameStr.split(‘.‘)[0]     #take off .txt
80             classNumStr = int(fileStr.split(‘_‘)[0])
81             hwLabels.append(classNumStr)
82             trainingMat[i,:] = img2vector(‘trainingDigits/%s‘ % fileNameStr)
83         testFileList = listdir(‘testDigits‘)        #iterate through the test set
84         errorCount = 0.0
85         mTest = len(testFileList)
86         for i in range(mTest):
87             fileNameStr = testFileList[i]
88             fileStr = fileNameStr.split(‘.‘)[0]     #take off .txt
89             classNumStr = int(fileStr.split(‘_‘)[0])
90             vectorUnderTest = img2vector(‘testDigits/%s‘ % fileNameStr)
91             classifierResult = classify0(vectorUnderTest, trainingMat, hwLabels, 3)
92             print "the classifier came back with: %d, the real answer is: %d" % (classifierResult, classNumStr)
93             if (classifierResult != classNumStr): errorCount += 1.0
94         print "\nthe total number of errors is: %d" % errorCount
95         print "\nthe total error rate is: %f" % (errorCount/float(mTest))  

  第三步：通过命令行交互

（1）：先将上述代码保存为kNN.py

（2）：再在IDLE下的run菜单下run一下，将其生成python模块

（3）： import  kNN（因为上一步已经生成knn模块）
      （4）： kNN.classify0([0,0],group,labels,3) （讨论[0,0]点属于哪一个类）

注：其中【0,0】可以随意换

即【】内的坐标就是我们要判断的点的坐标：

>>> kNN.classify0([0,0],group,labels,3)
‘B‘
>>> kNN.classify0([0,1],group,labels,3)
‘B‘
>>> kNN.classify0([0.6,0.6],group,labels,3)
‘A‘