【Python学习笔记】11：带散点图的k-means算法

带散点图的k-means算法

先安装好matplotlib包，用pip安装就可以。

import math
import random
f=open("dataforkmeans - .txt")
a=[]
b=[]
tag=1  
for k in f.read().split():  
    if tag==1:  
        a.append(float(k))  
        tag=0  
    else:  
        b.append(float(k))  
        tag=1  
#到此已经将x坐标和y坐标分别放在了a[]和b[]中  
  
p=len(a)  
k=4  
print ("数据表的长度为",p,"已知聚类数k =",k)
center=[]  
for i in range(k):  
    while(1):  
        j=random.randint(0,p-1)  
        if j not in center:  
            center+=[j]  
            break  
#到此已经选取了四个0~104的不重复下标  
cen_x=[a[i] for i in center]#存放种子点横坐标  
cen_y=[b[i] for i in center]#存放种子点纵坐标  
  
belong=[-1 for i in range(p)]  
#print belong  
tag=1  
count=0  
while(tag==1):#tag==1表示在循环里进行过种子点的移动  
    tag=0  
    #先更新聚类belong[]  
    for i in range(p):#对每一个点  
        index_maxlen=-1#存放到谁最近  
        maxlen=-1#存放最近距离  
        for j in range(k):#对k个种子点  
            numlen=(a[i]-cen_x[j])**2+(b[i]-cen_y[j])**2#存放当前距离  
            if maxlen<0 or numlen<maxlen:#如果比maxlen小  
                index_maxlen=j#记录种子下标  
                maxlen=numlen#更新更小值  
        belong[i]=index_maxlen#记录它的种子点下标  
    #再移动种子点  
    sum_x=[0 for i in range(4)]  
    sum_y=[0 for i in range(4)]#分别保存k个种子点对应聚类的横纵坐标和  
    numofclass=[0 for i in range(4)]#用来保存k个聚类的点数  
    for i in range(p):#对每一个点  
        sum_x[belong[i]]+=a[i]  
        sum_y[belong[i]]+=b[i]#将横纵坐标加到它的聚类中去  
        numofclass[belong[i]]+=1#维护聚类的点数  
    for i in range(k):#对k个种子点  
        if sum_x[i]/numofclass[i] !=cen_x[i]:  
            tag=1  
            cen_x[i]=sum_x[i]/numofclass[i]  
        if sum_y[i]/numofclass[i] !=cen_y[i]:  
            tag=1  
            cen_y[i]=sum_y[i]/numofclass[i]  
    count+=1  


x=[[],[],[],[]]
y=[[],[],[],[]]
for i in range(k):#对k个种子点  
        for j in range(p):#对每个点  
            if belong[j]==i:#如果是第i个聚类  
                x[i].append(a[j])
                y[i].append(b[j])
import numpy
import matplotlib
import pylab as pb
for i in range(k):
    if i==0:
        pb.plot(x[i],y[i],'or')
    elif i==1:
        pb.plot(x[i],y[i],'ob')
    elif i==2:
        pb.plot(x[i],y[i],'og')
    else:
        pb.plot(x[i],y[i],'ok')
pb.show()

运行结果：