Kmeans算法在python的实现--简易二类区分

本代码可以通过图像展现出聚合结果，帮助理解。

import random

import sys

import matplotlib.pyplot as plt

#K均值聚类法

def randList(size):

    all_points = []

    for i in range(size):

      datas = [random.randint(1, 100), random.randint(1, 100)]

      if not datas in all_points:  # 去掉重复数据

            all_points.append(datas)

    print(all_points)

    return all_points

#最简单的二类区分 需要不断迭代过程

def  Kmeans(AtypeList,BtypeList,randCenterA,randCenterB,initList,counts):

    lastAtypeList = AtypeList

    lastBtypeList = BtypeList

    AtypeList=[]

    BtypeList=[]

    for iL in initList:

        distanceToA = ((randCenterA[0] - iL[0]) * (randCenterA[0] - iL[0]) + (randCenterA[1] - iL[1]) * (

        randCenterA[1] - iL[1])) ** (0.5)

        distanceToB = ((randCenterB[0] - iL[0]) * (randCenterB[0] - iL[0]) + (randCenterA[1] - iL[1]) * (

        randCenterA[1] - iL[1])) ** (0.5)

        if distanceToA > distanceToB:

            AtypeList.append(iL)

        else:

            BtypeList.append(iL)

    #求得各类元素数量：

    Anum = len(AtypeList)

    Bnum = len(BtypeList)

    newAxSum=0

    newAySum=0

    newBxSum=0

    newBySum=0

    for lA in AtypeList:

        newAxSum=newAxSum+lA[0]

        newAySum=newAySum+lA[1]

    for lB in BtypeList:

        newBxSum = newBxSum + lB[0]

        newBySum = newBySum + lB[1]

    randCenterA=[newAxSum/Anum,newAySum/Anum]

    randCenterB=[newBxSum/Bnum,newBySum/Bnum]

    #反复迭代，直至聚类元素不变为止

    if (lastAtypeList==AtypeList and lastBtypeList==BtypeList) or counts > 1000 :

        print('迭代结束')

        print('质心A为：'+str(randCenterA))

        print('质心B为：' + str(randCenterB))

        print('聚类A元素为：' + str(AtypeList))

        print('聚类B元素为：' + str(BtypeList))

        print('迭代次数：' + str(counts))

        #开始绘制图谱

        for Aty in AtypeList:

            plt.scatter(Aty[0],Aty[1],c='b')

        for Bty in BtypeList:

            plt.scatter(Bty[0],Bty[1],c='g')

        plt.scatter(randCenterA[0], randCenterA[1], c='r')

        plt.scatter(randCenterB[0], randCenterB[1], c='r')

        plt.show()

    else:

      counts=counts+1

      Kmeans(AtypeList,BtypeList,randCenterA,randCenterB,initList,counts)

def ExampleSloveAndPaint(size):

  initList = randList(size)

  x=0

  #初始聚类中心 不一样的情况下，聚合结果会有区别

  while x<1:

      print('原始数组为：' + str(initList))

      randCenterA = [random.randint(1, 100), random.randint(1, 100)]

      randCenterB = [random.randint(1, 100), random.randint(1, 100)]

      Kmeans([], [], randCenterA, randCenterB, initList, 0)

      x=x+1

def main ():

sys.setrecursionlimit(2000) #设置迭代上限

ExampleSloveAndPaint(20) #设置聚类数组的元素个数

main()