Kmeans的python代码实现

import math
import random
import pandas as pd

#计算距离
def distance(a,b):
    distan = math.sqrt((a[0] - b[0])**2+(a[1] - b[1])**2)
    return distan

#D：数据集，k：簇数，maxIter：最大迭代次数，：簇划分的集合，C：质心集合
def kmeans(D,k,maxIter):
    if(k>=len(D)):
        return D
    C = []
    U = []
    #初始化簇划分集合
    for i in range(k):
        U.append([])
    #从D中随机选择k个样本作为初始质心{u1,u2,...,uk}
    ramdomNum = random.sample(range(0,len(D)),k)
    for i in range(k):
        C.append(D[ramdomNum[i]])
    Iter = 0
    tag = 0
    temp = 0
    C1 = C
    #计算各点到各质心的距离，划入距离最小的质心的簇。
    while(Iter < maxIter):
        i = 0
        while (i < len(D)):
            j = 0
            distan = 1000
            while (j < k):
                temp = distance(D[i], C[j])
                if (temp < distan):
                    tag = j
                    distan = temp
                j = j + 1
            U[tag].append(D[i])
            i = i + 1
        # 在小于最大迭代次数时，找出新质心
        C.clear()
        Iter = Iter + 1
        if(Iter < maxIter):
            break
        for num in range(k):
            C.append(findNew(U[num]))
            U[num].clear()
        if(C1 == C & Iter >=5 ):
            break
        C1 = C
    return U

#找出新质心
def findNew(L):
    x = 0
    y = 0
    tup = ()
    for i in range(len(L)):
        x = x + L[i][0]
        y = y + L[i][1]
        tup = (x/len(L),y/len(L))
    return tup

if __name__ =="__main__":
    dataset = pd.read_csv('data.csv',delimiter=',')
    data = dataset.values.tolist()
    D = []
    for i in range(len(data)):
        D.append(tuple(data[i]))
    num = int(input("请输入簇数："));
    U = kmeans(D,num,10)
    print(U)

data.csv中内容

运行结果：