kNearestNeighbors实战2使用kNN算法改进约会网站的配对结果

本文《machine learning in action》学习笔记
数据源码可以在这里获取 &＃xff1a;https://www.manning.com/books/machine-learning-in-action

这里Python 3&＃43;的code

from numpy import *
import matplotlib.pyplot as plt
import operatordef kNNClassify(inX, dataSet, labels, k):&＃39;&＃39;&＃39;put the kNN classification algorithm into action&＃39;&＃39;&＃39;dataSetSize &＃61; dataSet.shape[0]diffMax &＃61; tile(inX,(dataSetSize,1)) - dataSetsqDiffMax &＃61; diffMax ** 2sqDistances &＃61; sqDiffMax.sum(axis&＃61;1)distances &＃61; sqDistances**0.5# argsort 返回由大到小的索引值sortedDistIndicies &＃61; distances.argsort()classCount&＃61; {}for i in range(k):# 找到最大索引值对应数据的labelvoteIlabel &＃61; labels[sortedDistIndicies[i]]# returns a value for the given keyclassCount[voteIlabel] &＃61; classCount.get(voteIlabel,0) &＃43; 1# 按照键值的大小排列sortedClassCount &＃61; sorted(classCount.items(), key &＃61; operator.itemgetter(1), reverse &＃61; True)return sortedClassCount[0][0]def file2matrix(filename):"""process the text information"""fr &＃61; open(filename)arrayofLines &＃61; fr.readlines()fr.close()numberofLines &＃61; len(arrayofLines)returnMat &＃61; zeros((numberofLines, 3))classLabelVector &＃61; []index &＃61; 0for line in arrayofLines:# Python strip() 方法用于移除字符串头尾指定的字符&＃xff08;默认为空格或换行符&＃xff09;line &＃61; line.strip()# split()通过指定分隔符对字符串进行切片&＃xff0c;这里使用tablistFromLine &＃61; line.split(&＃39;\t&＃39;)returnMat[index,:] &＃61; listFromLine[0:3]classLabelVector.append(int(listFromLine[-1]))index &＃43;&＃61; 1return returnMat, classLabelVectordef autonorm(dataset):"""归一化"""# 求每一列最小值和最大值minvalue &＃61; dataset.min(0)maxvalue &＃61; dataset.max(0)ranges &＃61; maxvalue - minvalue# 使用shape获取dataset的shapenormdataset &＃61; zeros(shape(dataset))# shape[0] 获取第一行元素个数m &＃61; dataset.shape[0]# 使用tile函数将变量内容复制成输入矩阵同样大小的矩阵做矩阵的减法normdataset &＃61; dataset - tile(minvalue, (m,1))# 做矩阵的除法normdataset &＃61; normdataset / tile(ranges, (m,1))return normdataset, ranges, minvaluedef datingclasstest():horatio &＃61; 0.1datingdatamat, datinglabel &＃61; file2matrix("datingTestset2.txt")normat, ranges, minvalues &＃61; autonorm(datingdatamat)m &＃61; normat.shape[0]numtestvec &＃61; int(m*horatio)errorcount &＃61; 0.0for i in range(numtestvec):#classifierresult &＃61; classify0(normat[i,:], normat[numtestvec:m,:], datinglabel[numtestvec:m,:], 3)classifierresult &＃61; kNNClassify(normat[i, :], normat[numtestvec:m], datinglabel[numtestvec:m], 3)print("No.%d test data, the classifier came back with : %d, the real answeris: %d" %(i, classifierresult, datinglabel[i]))if (classifierresult !&＃61; datinglabel[i]):errorcount &＃43;&＃61; 1.0print ("the total error rate is: %f" % (errorcount/float(numtestvec)))if __name__ &＃61;&＃61; &＃39;__main__&＃39;:datingdatamat, datinglabel &＃61; file2matrix(&＃39;datingTestSet2.txt&＃39;)normdataset, ranges, minvalue &＃61; autonorm(datingdatamat)print(normdataset)print("ranges &＃61; ", ranges)print("minvalue &＃61; ", minvalue)fig &＃61; plt.figure()ax &＃61; fig.add_subplot(111)# 画散点图 scatterax.scatter(datingdatamat[:,1], datingdatamat[:,2], 15.0*array(datinglabel), 15.0*array(datinglabel))plt.show()datingclasstest()

读取TXT数据

def file2matrix(filename):"""process the text information"""fr &＃61; open(filename)arrayofLines &＃61; fr.readlines()numberofLines &＃61; len(arrayofLines)returnMat &＃61; zeros((numberofLines, 3))classLabelVector &＃61; []index &＃61; 0for line in arrayofLines:# Python strip() 方法用于移除字符串头尾指定的字符&＃xff08;默认为空格或换行符&＃xff09;line &＃61; line.strip()# split()通过指定分隔符对字符串进行切片&＃xff0c;这里使用tablistFromLine &＃61; line.split(&＃39;\t&＃39;)returnMat[index:1] &＃61; listFromLine[0:3]classLabelVector.append(int(listFromLine[-1]))index &＃43;&＃61; 1return returnMat, classLabelVector

normalization result&＃xff1a;

plot&＃xff1a;

test result&＃xff1a;

结果与书中的结果并不一致&＃xff0c;kNN这么简单的算法&＃xff0c;其结果应该一致才对。为什么&＃xff1f;