knn进阶-约会网站

import tensorflow as tf
import matplotlib.pyplot as plt
#from tensorflow.examples.tutorials.mnist import input_data
import numpy as np
#from numpy import * #直接使用命名空间，后面不用使用.引用
import operator as op

num_label = {'didntLike':1, 'smallDoses':2, 'largeDoses':3}

def file2matrix(filename):
    fr = open(filename)
    arrayLine = fr.readlines() #返回包含全部行的列表
    Lines_num = len(arrayLine)
    returnMat = np.zeros((Lines_num, 3))
    classLabelVector = []
    index = 0
    for line in arrayLine:
        oneline = line.strip() #去掉头部和尾部指定字符，默认为换行符或者空格
        listFromLine = oneline.split('\t') #指定分隔符对字符串切片
        returnMat[index, :] = listFromLine[0:3]
        classLabelVector.append(num_label[listFromLine[-1]])
        index += 1
    return returnMat, classLabelVector


#数据归一化,分别计算每列最大最小值（即一行是一个case）
def toNorm(dataSet):
    minval = np.min(dataSet, axis=0)
    maxval = np.max(dataSet, axis=0)
    minmat = np.tile(minval, (dataSet.shape[0], 1))
    ranges = maxval-minval
    normmat = np.zeros(dataSet.shape)
    print(type(normmat))
    normmat = (dataSet-minmat)/ranges
    return normmat, ranges, minval


def classify(inX, dataSet, label, k):
    dataSetSize = dataSet.shape[0]
    diffMat = np.tile(inX, (dataSetSize, 1)) - dataSet
    sqdiffMat = diffMat**2 #只有在.dot才当做矩阵相乘
    sqDistance = sqdiffMat.sum(axis=1)
    distance = sqDistance**0.5
    sortdisindex = np.argsort(distance) #从小到大依次返回对应元素下标而非值
    classCount = {}
    for i in range(k):
        votelable = label[sortdisindex[i]]
        classCount[votelable] = classCount.get(votelable, 0)+1


    sortclassCount = sorted(classCount.items(), key=op.itemgetter(1), reverse=True)
    return sortclassCount[0][0] #sortclassCount （'A', 2）



returnMat, classLabelVector = file2matrix('datingTestSet.txt')
normmat, ranges, minval= toNorm(returnMat)
print(type(normmat))
testRatio = 0.1
numTest = int(testRatio*normmat.shape[0])
error = 0.0
for i in range(numTest):
    classifyRes = classify(normmat[i, :], normmat[numTest:-1], classLabelVector, 3)
    if classifyRes != classLabelVector[i] :
        error +=1.0

print(1 - error/numTest)