本文共 5378 字，大约阅读时间需要 17 分钟。

KNN算法是懒惰的学习算法，没有明显的训练过程，预测时只需要使用已经有标注（分类学习）的训练数据即可

适用于多分类的学习任务

from numpy import*import operatorimport pdbimport matplotlibimport matplotlib.pyplot as pltfrom os import listdir#测试数据def createDataSet():    group = array([[1.0,1.1],[1.0,1.0],[0,0],[0,0.1]])    labels = ['A','A','B','B']    return group,labelsdef classify0(inX, dataSet, labels, k):    '''    K-邻近算法：    inX:目标点    dataSet：数据集    labels:数据的标签，label的列数和dataSet一样    K：选取的K个邻近值    返回inx的属性    '''    dataSetSize = dataSet.shape[0]  # 数据集大小    # 目标点到k邻近点的距离    diffMat = tile(inX, (dataSetSize, 1)) - dataSet    sqDiffMat = diffMat ** 2  # 平方    sqDistances = sqDiffMat.sum(axis=1)  # 求和    distances = sqDistances ** 0.5  # 开方    sortedDistIndicies = distances.argsort()  # 距离排序，返回的是distance排序后的索引    # 统计K个点所属类别    classCount = {}    for i in range(k):        voteIlabel = labels[sortedDistIndicies[i]]        classCount[voteIlabel] = classCount.get(voteIlabel, 0) + 1    sortedClassCount = sorted(classCount.items(), key=operator.itemgetter(1), reverse=True)    # 返回频率最高的标签类别    return sortedClassCount[0][0]#将读取文件，将文件转化为numpy数据def file2matrix(filename):    fr = open(filename)    arrayOLines = fr.readlines()    numberOfLines = len(arrayOLines) #文件的行数    returnMat = zeros((numberOfLines,3)) #初始化矩阵#     pdb.set_trace()    classLabelVector = []    index = 0    for line in arrayOLines:        line = line.strip()        listFromLine = line.split('\t')        returnMat[index,:] = listFromLine[0:3]        classLabelVector.append(int(listFromLine[-1]))        index += 1    return returnMat,classLabelVector#归一化数据def autoNorm(dataSet):#     pdb.set_trace()    minVals = dataSet.min(0) #行最大，若参数为1 则是列最大    maxVals = dataSet.max(0)    ranges = maxVals - minVals    normDataSet = zeros(shape(dataSet))    m = dataSet.shape[0] #dataSet的行数，若参数为1则为列数    normDataSet = dataSet - tile(minVals,(m,1)) #计算每个数据与最小数据之间的差值    normDataSet = normDataSet/tile(ranges,(m,1))#归一化    return normDataSet,ranges,minVals#绘图def data_plt(datingDataMat,datingLabels):    fig = plt.figure()    ax = fig.add_subplot(111)    ax.scatter(datingDataMat[:, 1], datingDataMat[:, 2], 15.0 * array(datingLabels), 15.0 * array(datingLabels))    plt.show()#分类器针对约会网站的测试代码def datingClassText(datingDataMat,datingLabels):    hoRation = 0.10 #10%测试集    m = datingDataMat.shape[0]    numTestVecs = int(m*hoRation) #测试集总量    errorCount = 0.0    for i in range(numTestVecs):        '''        classify0参数        （预测点，数据集，标注，K的取值）        '''        classifierResult = classify0(datingDataMat[i,:],datingDataMat[numTestVecs:m,:],datingLabels[numTestVecs:m],3)        print(classifierResult,datingLabels[i])        if(classifierResult != datingLabels[i]):            errorCount += 1.0 #错误分类计数    print(errorCount/float(numTestVecs)) #错误分类概率#约会网站预测函数def classifyPerson(datingDataMat,datingLabels,ranges,minVals):    resultList = ['not at all','in small doses','in large doses']    #输入待测试数据，并将测试数据转换成numpy数据    percentTats = float(input("每年玩游戏的时间"))    ffMiles = float(input("每年飞行公里数"))    iceCream = float(input("每年冰激凌消耗数"))    inArr = array([ffMiles,percentTats,iceCream])    #使用KNN算法    classifierResult = classify0((inArr-minVals)/ranges,datingDataMat,datingLabels,3)    #输出预测结果    print('the result is :',resultList[classifierResult-1])'''手写识别系统数据解释：trainingDigits是训练数据，testDigits是测试数据'''#每次只读一张图片（一个txt文件），将每张图片合成的数据，改成1*1024的numpy矩阵def img2vector(filename):    returnVect = zeros((1,1024))    fr = open(filename)    for i in range(32):        lineStr = fr.readline()        for j in range(32):            returnVect[0,32*i+j] = int(lineStr[j])    return returnVect#识别程序def handwritingClassTest(traindir,testdir):    '''     读取训练数据，生成训练集    '''    hwLabels = [] #用于存储每张图片所表示的数字,即数据属性    trainingFileList = listdir(traindir) #读取指定文件下下所有文件    m = len(trainingFileList)#计算总文件数目    trainingMat = zeros((m,1024))#存储所有图片数字矩阵，m*1024，    for i in range(m):#"0_7.txt" 文件名样式        fileNameStr = trainingFileList[i] #第i个文件的文件名        fileStr = fileNameStr.split('.')[0] #取文件名的前面，即去掉txt        classNumStr = int(fileStr.split('_')[0]) #从文件名获取当前文件所保存的图像表示的数字        hwLabels.append(classNumStr)        path = '{}/{}'.format(traindir, fileNameStr)        trainingMat[i:] = img2vector(path) #循环读入每个文件    '''    读取测试数据，对每个数据进行测试    '''    testFileList = listdir(testdir)    errorCount = 0.0    mTest = len(testFileList)    for i in range(mTest):        fileNameStr = testFileList[i]        fileStr = fileNameStr.split('.')[0]        classNumStr = int(fileStr.split('_')[0])        vectorUnderTest = img2vector('{}/{}'.format(testdir,fileNameStr))        classifierResult = classify0(vectorUnderTest,trainingMat,hwLabels,3) #进行预测        if classifierResult != classNumStr: #错误预测结果计数            errorCount += 1        print(" {},   {}".format(classifierResult,classNumStr))    print('the error rate is:',errorCount/float(mTest)) #错误预测率if __name__ == '__main__':    # group,labels = createDataSet()    # datingDataMat,datingLabels = file2matrix('datingTestSet2.txt')    # datingDataMat,ranges,minVals = autoNorm(datingDataMat)    # # data_plt(datingDataMat,datingLabels)    # datingClassText(datingDataMat,datingLabels)    # classifyPerson(datingDataMat,datingLabels,ranges,minVals)    # img2vector('digits/trainingDigits/0_1.txt')    path_train = 'digits/trainingDigits'    path_test = 'digits/testDigits'    handwritingClassTest(path_train,path_test)    # pdb.set_trace()

转载地址：http://aenvb.baihongyu.com/