《机器学习实战》的学习笔记之KNN

这里是《机器学习实战》中第二章KNN的代码部分解释。

---

KNN最常用的是欧式距离，它没有训练过程，直接就是分类

常用的向量距离度量准则：

欧式距离、曼哈顿距离、切比雪夫距离、马氏距离、巴氏距离、汉明距离、皮尔逊系数、信息熵，部分相关公式与python代码见：

https://blog.csdn.net/weixin_43330946/article/details/105032182

优点：精度高、对异常值不敏感、无数据输入假定（朴素贝叶斯需要假设样本之间独立、高斯分布）。

缺点：计算复杂度高（每一个样本都要计算）、空间复杂度高。

使用数据范围：数值型和标称型。

代码1：

已知4个样本的类别，再输入一个新的样本判断其属于哪一类：

import numpy as np
import operator

def creatDataSet():
    group = np.array([[1,101], [5,89], [100,5], [115,8]])
    labels = ['爱情片','爱情片','动作片','动作片']
    return group, labels


def classify0(inX, dataSet, labels, k):
    dataSetSize = dataSet.shape[0]#0表示行数
    #np.tile表示复制：在列方向上重复inX共1次，行方向上重复inX共dataSetSize次
    diffMat = np.tile(inX, (dataSetSize, 1)) - dataSet
    sqDiffMat = diffMat ** 2#特征相减后平方
    sqDistances = sqDiffMat.sum(axis=1)#sum(0)列相加，sum(1)行相加
    distances = sqDistances ** 0.5
    sortedDistIndices = distances.argsort()#返回distance中元素从小到大排序后的索引值
    #定义一个记录类别次数的字典
    classCount = {}
    for i in range(k):
        #取出前k个样本的相关索引
        voteIlable = labels[sortedDistIndices[i]]#取出第i个样本的类别
        #计算类别次数
        classCount[voteIlabel] = classCount.get(voteIlabel, 0) + 1
        #对获取的类别数量进行排序
    sortedClassCount = sorted(classCount.items(), key=operator.itemgetter(1), reverse=True)
    #key=operator.itemgetter(1)根据字典的值进行排序，
    #key=operator.itemgetter(0)根据字典的键进行排序，
    #reverse=True降序排序字典
    return sortedClassCount[0][0]


if __name__ == '__main__':
    group, labels = createDataSet()
    test = [101, 20]
    test_class = classify0(test , group, labels, 3)
    print(test_class)

代码2：

约会网站配对效果判定

import numpy as np
import matplotlib.pyplot as plt

def file2matrix(filename):
    fr = open(filename)
    arrayOLines = fr.readlines()
    numberOfLines = len(arrayOLines)
    returMat = np.zeros((numberOFLines, 3))#3个特征
    classLableVector = []#返回的分类标签向量
    index = 0#行的索引值

    for line in arrayOLines:
        line = line.strip()#默认删除空白符(包括\n、\r、\t、'')
        listFromLine = line.split('\t')#将字符串根据'\t'分隔符进行切片
        returnMat[index,:] = listFromLine[0:3]#将数据前三列提取出来放入returnMat中
        #根据文本中标记的喜欢的程度进行分类
        if listFromLine[-1] == 'didntlike':
            classLabelVector.append(1)
        elif listFromLine[-1] == 'smallDoses':
            classLabelVector.append(2)
        elif listFromLine[-1] == 'largeDoses':
            classLabelVector.append(3)
        index += 1
    return returnMat, classLabelVector


def showdatas(datingDataMat, datingLabels):
    font = FontProperties(fname=r'sumsun.ttc', size=14)#设置汉字格式，英语字体可以不用
    fig, axs = plt.subplots(nrows=2, ncols=2, sharex=False, sharey=Flase, figsize=(13,8))

    LabelsColors = []
    for i in datingLabels:
        if i == 1:
            LabelsColors.append('black')
        if i == 2:
            LabelsColors.append('orange')
        if i == 3:
            LabelsColors.append('red')

    axs[0][0].scatter(x=datingDataMat[:,0], y=datingDataMat[:,1], color=LabelsColors, s=15, alpha=.5)#根据datingDataMat的第一、二列数据画散点数据
    axs0_title_text = axs[0][0].set_title(u'每年获得的飞行常客里程数与玩视频游戏所消耗时间占比', FontProperties=font)
    axs0_xlabel_text = axs[0][0].set_xlabel(u'每年获得的飞行常客里程数', FontProperties=font)
    axs0_ylabel_text = axs[0][0].set_ylabel(u'玩视频游戏所消耗时间占比', FontProperties=font)
    plt.setp(axs0_title_text, size=9, weight='bold', color='red')
    plt.setp(axs0_xlabel_text, size=7, weight='bold', color='black')
    plt.setp(axs0_ylabel_text, size=7, weight='bold', color='black')

    axs[0][1].scatter(x=datingDataMat[:,0], y=datingDataMat[:,2], color=LabelsColors, s=15, alpha=.5)#根据datingDataMat的第一、三列数据画散点数据
    axs1_title_text = axs[0][1].set_title(u'每年获得的飞行常客里程数与每周消费的冰淇淋公升数', FontProperties=font)
    axs1_xlabel_text = axs[0][1].set_xlabel(u'每年获得的飞行常客里程数', FontProperties=font)
    axs1_ylabel_text = axs[0][1].set_ylabel(u'每周消费的冰淇淋公升数', FontProperties=font)
    plt.setp(axs1_title_text, size=9, weight='bold', color='red')
    plt.setp(axs1_xlabel_text, size=7, weight='bold', color='black')
    plt.setp(axs1_ylabel_text, size=7, weight='bold', color='black')

    axs[1][0].scatter(x=datingDataMat[:,1], y=datingDataMat[:,2], color=LabelsColors, s=15, alpha=.5)#根据datingDataMat的第二、三列数据画散点数据
    axs2_title_text = axs[1][0].set_title(u'玩视频游戏所消耗时间占比与每周消费的冰淇淋公升数', FontProperties=font)
    axs2_xlabel_text = axs[1][0].set_xlabel(u'玩视频游戏所消耗时间占比', FontProperties=font)
    axs2_ylabel_text = axs[1][0].set_ylabel(u'每周消费的冰淇淋公升数', FontProperties=font)
    plt.setp(axs2_title_text, size=9, weight='bold', color='red')
    plt.setp(axs2_xlabel_text, size=7, weight='bold', color='black')
    plt.setp(axs2_ylabel_text, size=7, weight='bold', color='black')

    #设置图例    
    didntLike = mlines.line2D([], [], color='black', marker='.', markersize=6, label='didntLike')
    smallDoses = mlines.line2D([], [], color='orange', marker='.', markersize=6, label='smallDoses')
    largeDoses = mlines.line2D([], [], color='red', marker='.', markersize=6, label='largeDoses')

    #添加图例
    axs[0][0].legend(handles=[didntLike, smallDoses, largeDoses])
    axs[0][1].legend(handles=[didntLike, smallDoses, largeDoses])
    axs[1][0].legend(handles=[didntLike, smallDoses, largeDoses])

    plt.show()


def autoNorm(dataSet):
    minVals = dataSet.min(0)#返回每一列的最小数
    maxVals = dataSet.max(0)
    ranges = maxVals - minVals
    normDataSet = np.zeros(np.shape(dataSet))
    m = dataSet.shape[0]
    normDataSet = dataSet - np.tile(minVals, (m,1))
    normDataSet = normDataSet / np.tile(ranges, (m,1))
    return normDataSet, ranges, minVals


def datingClassTest():
    filename = 'datingTestSet.txt'
    datingDataMat, datingLabels = file2matrix(filename)
    hoRatio = 0.10#10%作为测试集
    normMat, ranges, minVals = autoNorm(datingDataMat)#数据归一化，返回归一化后的矩阵、数据范围、数据最小值
    m = normMat.shape[0]
    numTestVecs = int(m*hoRatio)
    erroeCount = 0.0

    for i in range(numTestVecs):
        classfierResult = classify0(normMat[i:], normMat[numTestVecs:m,:], datingLabels[numTestVecs:m], 4)
        print('分类结果：%d\t真实类别:%d' % (classifierResult, datingLabels[i]))
        if classifierResult != datingLabels[i]:
            errorCount += 1.0
    print('错误率:%f%%' % (errorCount / float(numTestVecs) * 100))
        


def classify0(inX, datsSet, labels, k):
    dataSetSize = dataSet.shape[0]
    diffMat = np.tile(inX, (dataSetSize, 1)) - dataSet
    sqDiffMat = diffMat ** 2
    sqDistances = sqDiffMat.sum(axis=1)
    distances = sqDistances ** 0.5
    sortedDistIndices = distances.argsort()
    classCount = {}

    for i in range(k):
        voteIlable = labels[sortedDistIndices[i]]#取出第i个样本的类别
        classCount[voteIlabel] = classCount.get(voteIlabel, 0) + 1
    sortedClassCount = sorted(classCount.items(), key=operator.itemgetter(1), reverse=True)

    return sortedClassCount[0][0]


def classifyPerson():
    resultlist = ['讨厌','有些喜欢','非常喜欢']
    percentTats = float(input('玩视频游戏所耗时间百分比:'))
    ffMiles = float(input('每年获得的飞行常客里程数:'))
    iceCream = float(input('每周消费的冰淇淋公升数'))
    filename = 'datingTestSet.txt'
    datingDataMat, datingLabels = file2matrix(filename)
    normMat, ranges, minVals = autoNorm(datingDataMat)
    inArr = np.array([percentTats, ffMiles, iceCream])
    norminArr = (inArr - minVals) / ranges
    classifierResult = classify0(norminArr, normMat, datingLabels, 3)
    print('你可能%s这个人' % (resultlist[classifierResult]))


if __name__ == '__main__':
    filename = 'datingTestSet.txt'
    datingDataMat, datingLabels = file2matrix(filename)
    showdatas(datingDataMat, datingLabels)
    datingClassTest()
    classifyPerson()

代码3：

类似于代码4。

代码4：

手写数字识别

http://archive.ocs.uci.edu/ml（加州大学欧文学院），有许多验证机器学习算法的数据集。

from os import listdir
import numpy as np
form sklearn.neighbors import KNeighborsClassifier as KNN

def img2vextor(filename):
    returnVect = np.zeros((1,1024))
    #KNN不能处理二维数据，没有位置信息，所以只能输入一维信息，也说明了CNN的强大
    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


def handwritingClassTest():
    hwLabels = []
    trainingFileList = list('traingDigits')
    m = len(trainingFileList)
    trainingMat = np.zeros((m, 1024))
    for i in range(m):
        fileNameStr = trainingFileList[i]
        classNumber = int(fileNameStr.split('_')[0]#获得分类的数字
        hwLabels.append(classNumber)
        trainingMat[i:] = ing2vector('trainingDigits/%s' % (fileNameStr))

    neigh = KNN(n_neighbors=3, algorithm='auto')
    neigh.fit(trainingMat, hwLabels)
    testFileList = listdir('testDigits')
    errorCount = 0.0
    mTest = len(testFileList)
    for i in range(mTest):
        fileNameStr = testFileList[i]
        classNumber = int(fileNameStr.split('_')[0]
        vectorUnderTest = img2vector('testDigits/%s' % (fileNameStr))
        classifierResult = neigh.predict(vectorUnderTest)
        print('分类返回结果为%d\t真实结果为%d' % (classifierResult, classNumber))
        if (classifierResult != classNumber):
            errrorCount += 1.0
    print('总共错了%d个数据\n错误率为%f%%' % (errorCount, erroeCount / mTest * 100))


if __name__ == '__main__':
    handwritingClassTest()