机器学习k—近邻算法实例

一、利用k——近邻算法实现约会对象类型分类。（python3.5.3 ubuntu16.04）

1、创建文件knn.py，里面添加代码：

from numpy import *
import operator
from os import listdir

#k-近邻算法, inX为用于分类的输入向量, 训练样本集dataSet, 标签向量labels, 选择最近邻的个数k
def classify0(inX, dataSet, labels, k):
    dataSetSize = dataSet.shape[0]    #读取矩阵dataSet的第一维度的长度
    diffMat = tile(inX, (dataSetSize, 1)) - dataSet     #复制矩阵
    sqDiffMat = diffMat**2
    sqDistances = sqDiffMat.sum(axis = 1)
    distances = sqDistances**0.5
    sortedDistIndicies = distances.argsort()    #从小到大依次排序
    classCount = {}
    for i in range(k):
        voteIlabel = labels[sortedDistIndicies[i]]     #选取与当前距离最小的k个点
        classCount[voteIlabel] = classCount.get(voteIlabel,0) + 1    #确定前k个最小元素所在的分类
    sortedClassCount = sorted(classCount.items(), key = operator.itemgetter(1), reverse = True)   #从大到小对元组进行排序
    return sortedClassCount[0][0]    #返回前k个点中出现频率最高的类别作为当前点的预测分类

2、在knn.py中添加代码：将文本文件转换成矩阵

#文件转换成矩阵
def file2matrix(filename):
    fr = open(filename)
    numberOfLines = len(fr.readlines())         #get the number of lines in the file
    returnMat = zeros((numberOfLines,3))        #prepare matrix to return
    classLabelVector = []                       #prepare labels return
    fr = open(filename)
    index = 0
    for line in fr.readlines():
        line = line.strip()     #截取所有的回车字符
        listFromLine = line.split('\t')    #使用tab字符\t将上一步得到的整行数据分割成一个元素列表
        returnMat[index,:] = listFromLine[0:3]    #取前3个元素，将它们存储到特征矩阵中
        classLabelVector.append(int(listFromLine[-1]))    #将列表的最后一列存储到向量classLabelVector中
        index += 1
    return returnMat,classLabelVector

利用散点图分析数据

import numpy as np
import knn
import matplotlib
import matplotlib.pyplot as plt
from array import array

datingDataMat, datingLabels = knn.file2matrix('datingTestSet2.txt')
fig = plt.figure()
ax = fig.add_subplot(111)
#ax.scatter(datingDataMat[:,1],datingDataMat[:,2],15.0*numpy.array(datingLabels),15.0*numpy.array(datingLabels))
datingLabels = np.array(datingLabels)
idx_1 = np.where(datingLabels==1)
p1 = ax.scatter(datingDataMat[idx_1,0],datingDataMat[idx_1,1], s=20, marker = 'o', c = 'r', label='Do Not Like')
idx_2 = np.where(datingLabels==2)
p2 = ax.scatter(datingDataMat[idx_2,0],datingDataMat[idx_2,1], s=10, marker = 'o', c = 'b', label='Liked in Small Doses')
idx_3 = np.where(datingLabels==3)
p3 = ax.scatter(datingDataMat[idx_3,0],datingDataMat[idx_3,1], s=30, marker = 'o', c = 'g', label='Liked in Large Doses')
#ax.legend(fontsize = 5)
plt.legend(loc = 'upper left', fontsize = 10)
#plt.xlim(-5,35)
plt.xlabel('Frequent-flier miles per year')
plt.ylabel('The percentage of time spent on playing video games(%)')
#plt.ylabel('The litres of ice cream consumed per week')

#plt.xlim(0, 25)
#plt.ylim(0.0, 2.0)
plt.savefig("examples_1.jpg")
plt.show()

散点图：

3、归一化数值

def autoNorm(dataSet):
    minVals = dataSet.min(0)    #将每列的最小值放到minVals中,参数0是使得函数可以从列中选取最小值
    maxVals = dataSet.max(0)
    ranges = maxVals - minVals
    normDataSet = zeros(shape(dataSet))
    m = dataSet.shape[0]
    normDataSet = dataSet - tile(minVals, (m,1))
    normDataSet = normDataSet/tile(ranges, (m,1))
    return normDataSet, ranges, minVals

4、测试代码

def datingClassTest():
    hoRatio = 0.10
    datingDataMat, datingLables = file2matrix('datingTestSet2.txt')
    normMat, ranges, minVals = autoNorm(datingDataMat)
    m = normMat.shape[0]
    numTestVecs = int(m * hoRatio)
    errorCount = 0.0
    for i in range(numTestVecs):
        classifierResult = classify0(normMat[i,:], normMat[numTestVecs:m,:], datingLables[numTestVecs:m],3)
        print("the classifier came back with: %d, the real answer is: %d" % (classifierResult, datingLables[i]))
        if (classifierResult != datingLables[i]): errorCount += 1.0
    print("the total error rate is: %f" % (errorCount/float(numTestVecs)))
    print(errorCount)

在终端中进入到knn.py文件所在文件夹，输入python进入到python环境

>>>import (knn)

>>>datingDataMat, datingLabels = knn.file2matrix('datingTestSet2.txt')      //datingTestSet2.txt文件在《机器学习实战》官网上下载数据

>>>knn.datingClassTest()    //测试代码

结果如下：

5、约会网站预测函数

#通过输入下面三个参量来判断这个人是不是需要约会的类型
def classifyPerson():
    resultList = ['not at all', 'in small doses', 'in large doses']    #三种类型分类
    percentTats = float(input("percentage of time spent playing vedio games?"))    #输入玩视频游戏时间百分比
    ffMiles = float(input("frequent flier miles earned per year?"))    #输入每年飞行的公里数
    iceCream = float(input("liters of ice cream consumed per year?"))     #输入每周消费的冰淇淋公升数
    datingDataMat, datingLables = file2matrix('datingTestSet2.txt')   #读取元数据,并将其转换成矩阵
    normMat, ranges, minVals = autoNorm(datingDataMat)     #归一化处理
    inArr = array([ffMiles, percentTats, iceCream])
    classifierResult = classify0((inArr-minVals)/ranges, normMat, datingLables, 3)
    print("you will probably like this person: %s" %(resultList[classifierResult-1]))

测试结果：

二、手写字识别系统

1、将图像转换为测试向量

#手写识别系统,将32*32的图像转换成1*1024的向量
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

2、手写数字识别系统的测试代码

#将数据输入到分类器,检测分类器的执行效果
#k近邻算法的缺点是不需要提前训练，导致计算量很大
def handwritingClassTest():
    hwLabels = []
    trainingFileList = listdir('trainingDigits')    #读取训练文件
    m = len(trainingFileList)    #训练文件长度
    trainingMat = zeros((m, 1024))    #创建m行1024列的训练矩阵，该矩阵每行数据存储一个图像
    for i in range(m):
        fileNameStr = trainingFileList[i]      #获取文件名
        fileStr = fileNameStr.split('.')[0]     #以.为分隔符取文件名
        classNumStr = int(fileStr.split('_')[0])     #以_为分隔符取文件名
        hwLabels.append(classNumStr)    #在hwLabels数组后面添加数字
        trainingMat[i,:] = img2vector('trainingDigits/%s' % fileNameStr)
    testFileList = listdir('testDigits')
    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('testDigits/%s' % fileNameStr)
        # 每个测试样本与1934个训练样本做距离计算，取前三个距离最小的值，并取其类别，类别频率出现最高的即为识别出来的数字值
        classfierResult = classify0(vectorUnderTest, trainingMat, hwLabels, 3)
        print("the classfier came back with: %d, the real answer is: %d" %(classfierResult, classNumStr))
        if (classfierResult != classNumStr): errorCount += 1.0
    print("\nthe total number of errors is: %d" %errorCount)
    print("\nthe total error rate is: %f" %(errorCount/float(mTest)))

测试结果：