K均值Kmeans聚类算法

Halcom

Kmeans聚类算法——K均值聚类算法
百度网盘链接：
视频链接：http://pan.baidu.com/s/1jIwYfky
具体链接在halcom.cn论坛，联系人QQ：3283892722
该论坛是一个学习交流平台，我会逐一的和大家分享学习。
欢迎大家录制视频，并提交给我，我来设置视频，你可在论坛进行打赏分享。
视频专用播放器：http://halcom.cn/forum.php?mod=viewthread&tid=258&extra=page%3D1

使用环境：Win7-32bit-Anaconda2-4.3.1-Windows-x86.exe具体的代码如下：Kmeans_function.py子函数文件：

1. # -*- coding: utf-8 -*-
   
2. """
   
3. Created on Sat Apr 08 22:19:29 2017
   
4. 
   
5. @author: ysw
   
6. """
   
7. from numpy import *  
   
8. #import time
   
9. import matplotlib.pyplot as plt
   
10. 
    
11. # calculate Euclidean distance  
    
12. def euclDistance(vector1, vector2):  
    
13.     return sqrt(sum(power(vector2 - vector1, 2)))  
    
14. 
    
15. # init centroids with random samples  
    
16. def initCentroids(dataSet, k):  
    
17.     numSamples, dim = dataSet.shape  
    
18.     centroids = zeros((k, dim))  
    
19.     for i in range(k):  
    
20.         index = int(random.uniform(0, numSamples))  
    
21.         centroids[i, :] = dataSet[index, :]  
    
22.     return centroids  
    
23.   
    
24. # k-means cluster  
    
25. def kmeans(dataSet, k):  
    
26.     numSamples = dataSet.shape[0]  
    
27.     # first column stores which cluster this sample belongs to,  
    
28.     # second column stores the error between this sample and its centroid  
    
29.     clusterAssment = mat(zeros((numSamples, 2)))  
    
30.     clusterChanged = True  
    
31.   
    
32.     ## step 1: init centroids  
    
33.     centroids = initCentroids(dataSet, k)  
    
34.   
    
35.     while clusterChanged:  
    
36.         clusterChanged = False  
    
37.         ## for each sample  
    
38.         for i in xrange(numSamples):  
    
39.             minDist  = 100000000.0  
    
40.             minIndex = 0  
    
41.             ## for each centroid  
    
42.             ## step 2: find the centroid who is closest  
    
43.             for j in range(k):  
    
44.                 distance = euclDistance(centroids[j, :], dataSet[i, :])  
    
45.                 if distance < minDist:  
    
46.                     minDist  = distance  
    
47.                     minIndex = j  
    
48.               
    
49.             ## step 3: update its cluster  
    
50.             if clusterAssment[i, 0] != minIndex:  
    
51.                 clusterChanged = True  
    
52.                 clusterAssment[i, :] = minIndex, minDist**2  
    
53.   
    
54.         ## step 4: update centroids  
    
55.         for j in range(k):  
    
56.             pointsInCluster = dataSet[nonzero(clusterAssment[:, 0].A == j)[0]]  
    
57.             centroids[j, :] = mean(pointsInCluster, axis = 0)  
    
58.   
    
59.     print 'Congratulations, cluster complete!'  
    
60.     return centroids, clusterAssment  
    
61.   
    
62. # show your cluster only available with 2-D data  
    
63. def showCluster(dataSet, k, centroids, clusterAssment):  
    
64.     numSamples, dim = dataSet.shape  
    
65.     if dim != 2:  
    
66.         print "Sorry! I can not draw because the dimension of your data is not 2!"  
    
67.         return 1  
    
68.   
    
69.     mark = ['or', 'ob', 'og', 'ok', '^r', '+r', 'sr', 'dr', '<r', 'pr']  
    
70.     if k > len(mark):  
    
71.         print "Sorry! Your k is too large!"  
    
72.         return 1  
    
73.   
    
74.     # draw all samples  
    
75.     for i in xrange(numSamples):  
    
76.         markIndex = int(clusterAssment[i, 0])  
    
77.         plt.plot(dataSet[i, 0], dataSet[i, 1], mark[markIndex])  
    
78.   
    
79.     mark = ['Dr', 'Db', 'Dg', 'Dk', '^b', '+b', 'sb', 'db', '<b', 'pb']  
    
80.     # draw the centroids  
    
81.     for i in range(k):  
    
82.         plt.plot(centroids[i, 0], centroids[i, 1], mark[i], markersize = 12)  
    
83.   
    
84.     plt.show()  

复制代码

Kmeans_main.py主函数程序：

1. from numpy import *  
   
2. #import time
   
3. #import matplotlib.pyplot as plt
   
4. import Kmeans_function
   
5. 
   
6. ## step 1: load data  
   
7. print "step 1: load data..."  
   
8. dataSet = []  
   
9. fileIn = open(r'C:\Users\ysw\Desktop\Python(x,y)2.7.10\testSet.txt')  
   
10. for line in fileIn.readlines():  
    
11.     lineArr = line.strip().split()  
    
12.     dataSet.append([float(lineArr[0]), float(lineArr[1])])  
    
13.   
    
14. ## step 2: clustering...  
    
15. print "step 2: clustering..."  
    
16. dataSet = mat(dataSet)  
    
17. k = 4
    
18. centroids, clusterAssment = Kmeans_function.kmeans(dataSet, k)  
    
19. 
    
20. ## step 3: show the result  
    
21. print "step 3: show the result..."  
    
22. Kmeans_function.showCluster(dataSet, k, centroids, clusterAssment)  
    

复制代码

参考链接：http://blog.csdn.net/zouxy09/article/details/17589329
http://halcom.cn/forum.php?mod=viewthread&tid=2770&extra=