机器学习实战-学习笔记(一) kNN算法

kNN原理: 存在一个样本集合, 集合中每个样本都有标签; 输入一个不带标签的数据, 根据计算找出k个和这个数据最相近的样本, 这k个样本中出现最多次的标签, 作为这个数据的标签.

以下代码保存与kNN.py

from numpy import *
import operator

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

def classify0(inX, dataSet, labels, k):
"""Return the label that happens most.
Parameters:
	inX(Tuple): The data that needs to be classified.
	dataSet(Array): The data set that are used to train data.
	lables(List): List of labels contained in dataSet
	k:The number of chosen nearest neighbours.
"""
     #get the number of rows
    dataSetSize = dataSet.shape[0]
    #copy the data vertically to same size of dataSet and subtract it
    diffMat = tile(inX, (dataSetSize, 1)) - dataSet  
    #calculate the Euclidean distance
    sqDiffMat = diffMat ** 2 
    sqDistances = sqDiffMat.sum(axis=1)  
    distances = sqDistances ** 0.5  
    #get the indices(array) that sort the distaces array
    sortedDistIndicies = distances.argsort()  
    classCount = {}  
    for i in range(k):
    	#get the labels of first k nearest neighbours in dataSet
        voteIlabel = labels[sortedDistIndicies[i]] 
        #record the number of occurances for each label
        classCount[voteIlabel] = classCount.get(voteIlabel, 0) + 1
    #sort the items according to their values and store them in a list
    sortedClassCount = sorted(classCount.items(), key=operator.itemgetter(1), reverse=True)
    return sortedClassCount[0][0]

执行代码

>>> import kNN
>>> group, labels = kNN.createDataSet()
>>> kNN.classify0([0,0], group, labels, 3)
>'B'