机器学习之K近邻

K近邻算法在机器学习算法中是最容易理解并且是最容易使用的算法，下面是机器学习实战中对K近邻算法的注释。

'''
Created on Sep 16, 2010
kNN: k Nearest Neighbors

Input:      inX: vector to compare to existing dataset (1xN)
            dataSet: size m data set of known vectors (NxM)
            labels: data set labels (1xM vector)
            k: number of neighbors to use for comparison (should be an odd number)

Output:     the most popular class label

@author: pbharrin
'''
from numpy import *
import operator
from os import listdir

def classify0(inX, dataSet, labels, k):
    dataSetSize = dataSet.shape[0] #shape函数读取numpy矩阵或者数组的指定维度长度
    diffMat = tile(inX, (dataSetSize,1)) - dataSet#tile函数重复某个数组，构成一个新的数组
    sqDiffMat = diffMat**2#每个元素进行平方
    sqDistances = sqDiffMat.sum(axis=1)#sum按指定维度求和
    distances = sqDistances**0.5#每个元素开方
    sortedDistIndicies = distances.argsort()#argsort排序后得到原来数据的下标     
    classCount={} #定义一个字典         
    for i in range(k):
        voteIlabel = labels[sortedDistIndicies[i]]#返回指定下标的标签值
        classCount[voteIlabel] = classCount.get(voteIlabel,0) + 1#更新字典，对指定key值对应的值加1操作

        #sorted函数用来排列字典，items返回字典的元组对的列表，key是用列表的某个属性作为关键字，reverse指定是否逆序排列
    sortedClassCount = sorted(classCount.items(), key=operator.itemgetter(1), reverse=True)
    return sortedClassCount[0][0]

#创建数据集，返回数组
def createDataSet():
    group = array([[1.0,1.1],[1.0,1.0],[0,0],[0,0.1]])
    labels = ['A','A','B','B']
    return group, labels

#从文件中解析数据
def file2matrix(filename):
    fr = open(filename)#根据filename指定的文件地址打开文件
    numberOfLines = len(fr.readlines()) #以行的形式读取整个文件
    returnMat = zeros((numberOfLines,3)) #分配空间用来接下来存储特征向量
    classLabelVector = []               #分配空间用来接下来存储特征值对应的类别 
    fr = open(filename)#再一次打开指定文件
    index = 0
    for line in fr.readlines():#以行为单位遍历整个文档
        line = line.strip()#用于移除字符串头尾指定的字符（默认为空格）
        listFromLine = line.split('\t')#通过指定分隔符对字符串进行切片
        returnMat[index,:] = listFromLine[0:3]#将在3维空间每个样本的特征向量存入returnMat
        classLabelVector.append(int(listFromLine[-1]))#将listFromLine最后一维的值存入classLabelVector
        index += 1#更新index
    return returnMat,classLabelVector


def drawing(fileMat,classLabels):
    import matplotlib.pyplot as plt#导入绘图库
    fig=plt.figure()#新建一个figure对象
    ax=fig.add_subplot(111)#增加一个子窗口到fig
    ax.scatter(fileMat[:,1],fileMat[:,2],15.0*array(classLabels),15.0*array(classLabels))#在指定窗口绘制散点图
    plt.show()#显示窗口

#数值归一化    
def autoNorm(dataSet):
    minVals = dataSet.min(0)
    maxVals = dataSet.max(0)
    ranges = maxVals - minVals#得到最大值和最小值的差
    normDataSet = zeros(shape(dataSet))#shape返回多维数组最长维度，初始化normDataSet容器
    m = dataSet.shape[0]
    normDataSet = dataSet - tile(minVals, (m,1))#数组每个元素减去最小值
    normDataSet = normDataSet/tile(ranges, (m,1))   #得到最终归一化结果
    return normDataSet, ranges, minVals#返回归一化结果

def datingClassTest():
    hoRatio = 0.50      #hold out 10%
    datingDataMat,datingLabels = file2matrix('datingTestSet2.txt') #载入特征矩阵和特征值
    normMat, ranges, minVals = autoNorm(datingDataMat)#返回归一化后的特征矩阵，区间和最小值
    m = normMat.shape[0]
    numTestVecs = int(m*hoRatio)
    errorCount = 0.0
    for i in range(numTestVecs):#取前0.5的数据进行测试
        classifierResult = classify0(normMat[i,:],normMat[numTestVecs:m,:],datingLabels[numTestVecs:m],3)
        print(("the classifier came back with: %d, the real answer is: %d") % (classifierResult, datingLabels[i]))
        if (classifierResult != datingLabels[i]):
            errorCount += 1.0
    print(("the total error rate is: %f" )%(errorCount/float(numTestVecs)))
    print(errorCount)