K-近邻法

一、简易模型代码实现

import numpy as np
import operator

def creatDataset():
    group = np.array([[1,1.1],[1,1],[0,0],[0,0.1]])
    labels = ['A','A','B','B']
    return group,labels

def classify0(inX,dataSet,labels,k):
    '''
    k-近领算法
    
    ------------
    参数：
    inX：输入向量
    dataSet：输入样本集
    labels：标签向量
    k：选择最近邻的数目
    
    ------------
    返回：
    标签向量
    
    '''
    dataSetSize = dataSet.shape[0]
    difMat = np.tile(inX,(dataSetSize,1)) - dataSet
    sqDiffMat = difMat**2
    sqDistances = sqDiffMat.sum(axis=1)
    distances = sqDistances**0.5
    sortedDistIndecies = distances.argsort()
    classCount = {}
    for i in range(k):
        voteIlabel = labels[sortedDistIndecies[i]]
        classCount[voteIlabel] = classCount.get(voteIlabel,0)+1
    sortedClassCount = sorted(classCount.items(),key=operator.itemgetter(1),\
                                 reverse=True)
    return sortedClassCount[0][0]

dataSet,labels = creatDataset()

l = classify0(np.array([[1,2]]),dataSet,labels,3)

二、sklearn自带knn模型

'''
虹膜数据集是经典且非常容易的多类别分类数据集
'''
iris = datasets.load_iris(return_X_y = False)

'''
K-n邻居分类KNeighborsClassifier 是最常用的技术。数值的最佳选择K高度依赖数据：通常情况下，  
 K抑制了噪声的影响，但使分类边界不那么明显
'''
knn=neighbors.KNeighborsClassifier()

knn.fit(iris.data,iris.target)

predictedLabel=knn.predict([[0.8, 1, 0.3, 0.5]])

print('predictedLabel is:%s' %predictedLabel )

 

 

 三、使用K近邻算法改进约会网站的配对效果

1、准备数据

'''
准备数据
'''
def file2matrix(filename):
    fr = open(filename)
    #readlines() 方法用于读取所有行(直到结束符 EOF)并返回列表
    numberOfLines = len(fr.readlines())         
    returnMat = np.zeros((numberOfLines,3))        
    classLabelVector = []                         
    index = 0
    fr = open(filename)
    for line in fr.readlines():
        line = line.strip()                             #去掉每行头尾空白
        listFromLine = line.split('\t')              #split() 通过指定分隔符对字符串进行切片
        returnMat[index,:] = listFromLine[0:3]
        classLabelVector.append(int(listFromLine[-1]))
        index += 1
    fr.close()
    return returnMat,classLabelVector

datingDataMat,datingLabels = file2matrix('D:\spyder_code\机器学习实战\K-最近邻算法\datingTestSet2.txt')

2、数据处理

'''
不同列的差异过大，归一化处理
'''
def autoNorm(dataSet):
    minVals = dataSet.min(0)
    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))   #element wise divide
    return normDataSet, ranges, minVals

normDataSet, ranges, minVals = autoNorm(datingDataMat)

3、可视化

'''
绘图
'''
def plotfig(normDataSet,datingLabels):
    fig = plt.figure()
    ax = fig.add_subplot(111, projection='3d')
    
    data = np.insert(normDataSet,3,np.array(datingLabels),axis=1)
    for m,i in [('o',1),('^',2),('*',3)]:
        dt = data[data[:,3]==i]
        xs = dt[:,0]
        ys = dt[:,1]
        zs = dt[:,2]
        ax.scatter(xs, ys, zs, marker=m)
    
    ax.set_xlabel('X Label')
    ax.set_ylabel('Y Label')
    ax.set_zlabel('Z Label')
    
    plt.show()

plotfig(normDataSet,datingLabels)

 4、测试算法

'''
验证分类器
'''
def datingClassTest():
    hoRatio = 0.10      #hold out 10%
    datingDataMat,datingLabels = file2matrix('D:\spyder_code\机器学习实战\K-最近邻算法\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,:],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)))


datingClassTest()