聚类算法 实例

testSet.txt

1.658985 4.285136
-3.453687 3.424321
4.838138 -1.151539
-5.379713 -3.362104
0.972564 2.924086
-3.567919 1.531611
0.450614 -3.302219
-3.487105 -1.724432
2.668759 1.594842
-3.156485 3.191137
3.165506 -3.999838
-2.786837 -3.099354
4.208187 2.984927
-2.123337 2.943366
0.704199 -0.479481
-0.392370 -3.963704
2.831667 1.574018
-0.790153 3.343144
2.943496 -3.357075
-3.195883 -2.283926
2.336445 2.875106
-1.786345 2.554248
2.190101 -1.906020
-3.403367 -2.778288
1.778124 3.880832
-1.688346 2.230267
2.592976 -2.054368
-4.007257 -3.207066
2.257734 3.387564
-2.679011 0.785119
0.939512 -4.023563
-3.674424 -2.261084
2.046259 2.735279
-3.189470 1.780269
4.372646 -0.822248
-2.579316 -3.497576
1.889034 5.190400
-0.798747 2.185588
2.836520 -2.658556
-3.837877 -3.253815

places.txt

Dolphin II 10860 SW Beaverton-Hillsdale Hwy Beaverton, OR 45.486502 -122.788346
Hotties 10140 SW Canyon Rd. Beaverton, OR 45.493150 -122.781021
Pussycats 8666a SW Canyon Road Beaverton, OR 45.498187 -122.766147
Stars Cabaret 4570 Lombard Ave Beaverton, OR 45.485943 -122.800311
Sunset Strip 10205 SW Park Way Beaverton, OR 45.508203 -122.781853
Vegas VIP Room 10018 SW Canyon Rd Beaverton, OR 45.493398 -122.779628
Full Moon Bar and Grill 28014 Southeast Wally Road Boring, OR 45.430319 -122.376304
505 Club 505 Burnside Rd Gresham, OR 45.507621 -122.425553
Dolphin 17180 McLoughlin Blvd Milwaukie, OR 45.399070 -122.618893
Dolphin III 13305 SE McLoughlin BLVD Milwaukie, OR 45.427072 -122.634159
Acropolis 8325 McLoughlin Blvd Portland, OR 45.462173 -122.638846
Blush 5145 SE McLoughlin Blvd Portland, OR 45.485396 -122.646587
Boom Boom Room 8345 Barbur Blvd Portland, OR 45.464826 -122.699212
Bottoms Up 16900 Saint Helens Rd Portland, OR 45.646831 -122.842918
Cabaret II 17544 Stark St Portland, OR 45.519142 -122.482480
Cabaret Lounge 503 W Burnside Portland, OR 45.523094 -122.675528
Carnaval 330 SW 3rd Avenue Portland, OR 45.520682 -122.674206
Casa Diablo 2839 NW St. Helens Road Portland, OR 45.543016 -122.720828
Chantilly Lace 6723 Killingsworth St Portland, OR 45.562715 -122.593078
Club 205 9939 Stark St Portland, OR 45.519052 -122.561510
Club Rouge 403 SW Stark Portland, OR 45.520561 -122.675605
Dancin’ Bare 8440 Interstate Ave Portland, OR 45.584124 -122.682725
Devil’s Point 5305 SE Foster Rd Portland, OR 45.495365 -122.608366
Double Dribble 13550 Southeast Powell Boulevard Portland, OR 45.497750 -122.524073
Dream on Saloon 15920 Stark St Portland, OR 45.519142 -122.499672
DV8 5003 Powell Blvd Portland, OR 45.497498 -122.611177
Exotica 240 Columbia Blvd Portland, OR 45.583048 -122.668350
Frolics 8845 Sandy Blvd Portland, OR 45.555384 -122.571475
G-Spot Airport 8654 Sandy Blvd Portland, OR 45.554263 -122.574167
G-Spot Northeast 3400 NE 82nd Ave Portland, OR 45.547229 -122.578746
G-Spot Southeast 5241 SE 72nd Ave Portland, OR 45.484823 -122.589208
Glimmers 3532 Powell Blvd Portland, OR 45.496918 -122.627920
Golden Dragon Exotic Club 324 SW 3rd Ave Portland, OR 45.520714 -122.674189
Heat 12131 SE Holgate Blvd. Portland, OR 45.489637 -122.538196
Honeysuckle’s Lingerie 3520 82nd Ave Portland, OR 45.548651 -122.578730
Hush Playhouse 13560 Powell Blvd Portland, OR 45.497765 -122.523985
JD’s Bar & Grill 4523 NE 60th Ave Portland, OR 45.555811 -122.600881
Jody’s Bar And Grill 12035 Glisan St Portland, OR 45.526306 -122.538833
Landing Strip 6210 Columbia Blvd Portland, OR 45.595042 -122.728825
Lucky Devil Lounge 633 SE Powell Blvd Portland, OR 45.501585 -122.659310

#-*- coding: utf-8 -*- 

'''
Created on Feb 16, 2011
k Means Clustering for Ch10 of Machine Learning in Action
@author: Peter Harrington
'''
from numpy import *

#读数据
def loadDataSet(fileName):        
    dataMat = []   #创建列表，存储读取的数据
    fr = open(fileName)
    for line in fr.readlines(): #读每一行
        line1=line.strip();     #删头尾空白
        curLine = line1.split('\t') #以\t为分割，返回一个list列表
        fltLine = map(float,curLine)#str 转成  float
        dataMat.append(fltLine)     #将元素添加到列表尾
    return dataMat 

#算距离
def distEclud(vecA, vecB):                  #两个向量间欧式距离
    return sqrt(sum(power(vecA - vecB, 2))) #la.norm(vecA-vecB)

#初始化聚类中心
def randCent(dataSet, k):
    #特征维度
    n = shape(dataSet)[1]   
    #创建聚类中心的矩阵 k x n                
    centroids = mat(zeros((k,n)))  
    #遍历n维特征         
    for j in range(n):     
        #第j维特征属性值min   ,1x1矩阵                 
        minJ = min(dataSet[:,j])        
        #区间值max-min，float数值    
        rangeJ = float(max(dataSet[:,j]) - minJ)   
        #第j维，每次随机生成k个中心 
        centroids[:,j] = mat(minJ + rangeJ * random.rand(k,1))
    return centroids

#k-means算法  （#默认欧式距离，初始中心点方法randCent()）  
def kMeans(dataSet, k, distMeas=distEclud, createCent=randCent): 
    m = shape(dataSet)[0]   #样本总数
    #分配样本到最近的簇：存[簇序号,距离的平方] 
    clusterAssment = mat(zeros((m,2)))       
    #step1:#初始化聚类中心                                     
    centroids = createCent(dataSet, k)   
    clusterChanged = True
    #所有样本分配结果不再改变，迭代终止
    while clusterChanged:   
        clusterChanged = False        
        #step2:分配到最近的聚类中心对应的簇中
        for i in range(m):   
            minDist = inf; minIndex = -1  #对于每个样本，定义最小距离
            for j in range(k):  #计算每个样本与k个中心点距离
                distJI = distMeas(centroids[j,:],dataSet[i,:]) 
                if distJI < minDist: 
                    minDist = distJI; minIndex = j  #获取最小距离,及对应的簇序号
            if clusterAssment[i,0] != minIndex: clusterChanged = True 
            clusterAssment[i,:] = minIndex,minDist**2 #分配样本到最近的簇
        print 'centroids=',centroids        
        #step3:更新聚类中心
        for cent in range(k):#样本分配结束后，重新计算聚类中心
            #获取该簇所有的样本点
            ptsInClust = dataSet[nonzero(clusterAssment[:,0].A==cent)[0]]
            #更新聚类中心：axis=0沿列方向求均值
            centroids[cent,:] = mean(ptsInClust, axis=0) 
    return centroids, clusterAssment

#二分kmeans        
def biKmeans(dataSet, k, distMeas=distEclud):
    m = shape(dataSet)[0]
    clusterAssment = mat(zeros((m,2)))
    #所有样本看成一个簇，求均值
    centroid0 = mean(dataSet, axis=0).tolist()[0]#axis=0按列，matrix->list
    centList =[centroid0] #create a list with one centroid
    for j in range(m): #计算初始总误差SSE
        clusterAssment[j,1] = distMeas(mat(centroid0), dataSet[j,:])**2
    #当簇数<k时
    while (len(centList) < k):
        lowestSSE = inf  #初始化SSE
        for i in range(len(centList)):        #对每个簇
            #获取当前簇cluster=i内的数据
            ptsInCurrCluster = dataSet[nonzero(clusterAssment[:,0].A==i)[0],:]            
            #对cluster=i的簇进行kmeans划分，k=2
            centroidMat, splitClustAss = kMeans(ptsInCurrCluster, 2, distMeas)            
            #cluster=i的簇被划分为两个子簇后的SSE
            sseSplit = sum(splitClustAss[:,1])            
            #除了cluster=i的簇，其他簇的SSE
            sseNotSplit = sum(clusterAssment[nonzero(clusterAssment[:,0].A!=i)[0],1])
            print "sseSplit, and notSplit: ",sseSplit,sseNotSplit           
            #找最佳的划分簇，使得划分后 总SSE=sseSplit + sseNotSplit最小
            if (sseSplit + sseNotSplit) < lowestSSE: 
                bestCentToSplit = i    
                bestNewCents = centroidMat #被划分簇的两个新中心
                bestClustAss = splitClustAss.copy() #被划分簇的聚类结果0,1 ，及簇内SSE
                lowestSSE = sseSplit + sseNotSplit                
        #将最佳被划分簇的聚类结果为1的类别，更换类别为len(centList)
        bestClustAss[nonzero(bestClustAss[:,0].A == 1)[0],0] = len(centList)          
        #将最佳被划分簇的聚类结果为0的类别，更换类别为bestCentToSplit
        bestClustAss[nonzero(bestClustAss[:,0].A == 0)[0],0] = bestCentToSplit
        print 'the bestCentToSplit is: ',bestCentToSplit
        print 'the len of bestClustAss is: ', len(bestClustAss)        
        #将被划分簇的一个中心，替换为划分后的两个中心
        centList[bestCentToSplit] = bestNewCents[0,:].tolist()[0] 
        centList.append(bestNewCents[1,:].tolist()[0])       
        #更新整体的聚类效果clusterAssment(类别，SSE)
        clusterAssment[nonzero(clusterAssment[:,0].A == bestCentToSplit)[0],:]= bestClustAss
        #kMeans.datashow(dataSet,len(centList),mat(centlist),clusterAssment)   
    return mat(centList), clusterAssment

#2维数据聚类效果显示
def datashow(dataSet,k,centroids,clusterAssment):  #二维空间显示聚类结果
    from matplotlib import pyplot as plt
    num,dim=shape(dataSet)  #样本数num ,维数dim

    if dim!=2:
        print 'sorry,the dimension of your dataset is not 2!'
        return 1

    marksamples=['or','ob','og','ok','^r','sb','<g'] #样本图形标记
    if k>len(marksamples):
        print 'sorry,your k is too large,please add length of the marksample!'
        return 1

    #绘所有样本
    for i in range(num):
        markindex=int(clusterAssment[i,0])#矩阵形式转为int值, 簇序号
        #特征维对应坐标轴x,y；样本图形标记及大小
        plt.plot(dataSet[i,0],dataSet[i,1],marksamples[markindex],markersize=6)

    #绘中心点            
    markcentroids=['dr','db','dg','dk','^b','sk','<r']#聚类中心图形标记
    for i in range(k):
        plt.plot(centroids[i,0],centroids[i,1],markcentroids[i],markersize=15)

    plt.title('k-means cluster result') #标题        
    plt.show()

#2维原始数据显示
def datashow0(dataSet):  #二维空间显示聚类结果
    from matplotlib import pyplot as plt
    num,dim=shape(dataSet)  #样本数num ,维数dim

    if dim!=2:
        print 'sorry,the dimension of your dataset is not 2!'
        return 1

    marksamples=['or','ob','og','ok','^r','sb','<g'] #样本图形标记
    if k>len(marksamples):
        print 'sorry,your k is too large,please add length of the marksample!'
        return 1

    #绘所有样本
    for i in range(num):
        markindex=int(clusterAssment[i,0])#矩阵形式转为int值, 簇序号
        #特征维对应坐标轴x,y；样本图形标记及大小
        plt.plot(dataSet[i,0],dataSet[i,1],marksamples[markindex],markersize=6)

    #绘中心点            
    markcentroids=['dr','db','dg','dk','^b','sk','<r']#聚类中心图形标记
    for i in range(k):
        plt.plot(centroids[i,0],centroids[i,1],markcentroids[i],markersize=15)

    plt.title('dataset') #标题        
    plt.show()


import urllib
import json
def geoGrab(stAddress, city):
    apiStem = 'http://where.yahooapis.com/geocode?'  #create a dict and constants for the goecoder
    params = {}
    params['flags'] = 'J'#JSON return type
    params['appid'] = 'aaa0VN6k'
    params['location'] = '%s %s' % (stAddress, city)
    url_params = urllib.urlencode(params)
    yahooApi = apiStem + url_params      #print url_params
    print yahooApi
    c=urllib.urlopen(yahooApi)
    return json.loads(c.read())

from time import sleep
def massPlaceFind(fileName):
    fw = open('places.txt', 'w')
    for line in open(fileName).readlines():
        line = line.strip()
        lineArr = line.split('\t')
        retDict = geoGrab(lineArr[1], lineArr[2])
        if retDict['ResultSet']['Error'] == 0:
            lat = float(retDict['ResultSet']['Results'][0]['latitude'])
            lng = float(retDict['ResultSet']['Results'][0]['longitude'])
            print "%s\t%f\t%f" % (lineArr[0], lat, lng)
            fw.write('%s\t%f\t%f\n' % (line, lat, lng))
        else: print "error fetching"
        sleep(1)
    fw.close()

def distSLC(vecA, vecB):#Spherical Law of Cosines
    a = sin(vecA[0,1]*pi/180) * sin(vecB[0,1]*pi/180)
    b = cos(vecA[0,1]*pi/180) * cos(vecB[0,1]*pi/180) * \
                      cos(pi * (vecB[0,0]-vecA[0,0]) /180)
    return arccos(a + b)*6371.0 #pi is imported with numpy

import matplotlib
import matplotlib.pyplot as plt
def clusterClubs(numClust=5):
    datList = []
    for line in open('places.txt').readlines():
        lineArr = line.split('\t')
        datList.append([float(lineArr[4]), float(lineArr[3])])
    datMat = mat(datList)
    myCentroids, clustAssing = biKmeans(datMat, numClust, distMeas=distSLC)
    fig = plt.figure()
    rect=[0.1,0.1,0.8,0.8]
    scatterMarkers=['s', 'o', '^', '8', 'p', \
                    'd', 'v', 'h', '>', '<']
    axprops = dict(xticks=[], yticks=[])
    ax0=fig.add_axes(rect, label='ax0', **axprops)
    imgP = plt.imread('Portland.png')
    ax0.imshow(imgP)
    ax1=fig.add_axes(rect, label='ax1', frameon=False)
    for i in range(numClust):
        ptsInCurrCluster = datMat[nonzero(clustAssing[:,0].A==i)[0],:]
        markerStyle = scatterMarkers[i % len(scatterMarkers)]
        ax1.scatter(ptsInCurrCluster[:,0].flatten().A[0], ptsInCurrCluster[:,1].flatten().A[0], marker=markerStyle, s=90)
    ax1.scatter(myCentroids[:,0].flatten().A[0], myCentroids[:,1].flatten().A[0], marker='+', s=300)
    plt.show()

if __name__=='__main__':

    # #=====显示原始数据
#     # #获取样本数据
#     datamat=mat(loadDataSet('testSet.txt'))
#     #样本的个数和特征维数
#     num,dim=shape(datamat)
#     marksamples=['ok'] #样本图形标记
#     for i in range(num):
#         plt.plot(datamat[i,0],datamat[i,1],marksamples[0],markersize=6)
#     plt.title('dataset') #标题
#     plt.show()

### #=====kmeans聚类
##    k=4    #用户定义聚类数
##    # 获取样本数据
##    datamat=mat(loadDataSet('testSet.txt'))
##    run_num=8 #循环多次看多次的聚类效果
##    for i in range(run_num): #可循环多次看效果图
##        mycentroids,clusterAssment=kMeans(datamat,k)
##         # 绘图显示
##        datashow(datamat,k,mycentroids,clusterAssment)




 ###二分kmeans
     datamat2=mat(loadDataSet('testSet.txt'))
     k= 4
     for i in range(1):  #可以循环多次看效果图
         centlist,mynewassments=biKmeans(datamat2,k)
         datashow(datamat2,k,centlist,mynewassments)

#-*- coding: utf-8 -*- 
from numpy import*
from matplotlib import pyplot as plt
import kMeans
#####################################################
##每次划分都显示一下

#二分kmeans        
#def biKmeans(dataSet, k, distMeas=distEclud):
dataSet=mat(kMeans.loadDataSet('testSet.txt'))
k=4
distMeas=kMeans.distEclud
m = shape(dataSet)[0]
clusterAssment = mat(zeros((m,2)))
#所有样本看成一个簇，求均值
centroid0 = mean(dataSet, axis=0).tolist()[0]#axis=0按列，matrix->list
centList =[centroid0] #create a list with one centroid
for j in range(m): #计算初始总误差SSE
    clusterAssment[j,1] = distMeas(mat(centroid0), dataSet[j,:])**2
kMeans.datashow(dataSet,len(centList),mat(centList),clusterAssment) 
#当簇数<k时
while (len(centList) < k):
    lowestSSE = inf  #初始化SSE
    #对每个簇
    for i in range(len(centList)):
    #获取当前簇cluster=i内的数据
        ptsInCurrCluster = dataSet[nonzero(clusterAssment[:,0].A==i)[0],:]            
        #对cluster=i的簇进行kmeans划分，k=2
        centroidMat, splitClustAss = kMeans.kMeans(ptsInCurrCluster, 2, distMeas)          
        #cluster=i的簇被划分为两个子簇后的SSE
        sseSplit = sum(splitClustAss[:,1])            
        #除了cluster=i的簇，其他簇的SSE
        sseNotSplit = sum(clusterAssment[nonzero(clusterAssment[:,0].A!=i)[0],1])
        print "sseSplit, and notSplit: ",sseSplit,sseNotSplit            
        #找最佳的划分簇，使得划分后 总SSE=sseSplit + sseNotSplit最小
        if (sseSplit + sseNotSplit) < lowestSSE: 
            bestCentToSplit = i    
            bestNewCents = centroidMat #被划分簇的两个新中心
            bestClustAss = splitClustAss.copy() #被划分簇的聚类结果0,1 ，及簇内SSE
            lowestSSE = sseSplit + sseNotSplit                
    #将最佳被划分簇的聚类结果为1的类别，更换类别为len(centList)
    bestClustAss[nonzero(bestClustAss[:,0].A == 1)[0],0] = len(centList)          
    #将最佳被划分簇的聚类结果为0的类别，更换类别为bestCentToSplit
    bestClustAss[nonzero(bestClustAss[:,0].A == 0)[0],0] = bestCentToSplit
    print 'the bestCentToSplit is: ',bestCentToSplit
    print 'the len of bestClustAss is: ', len(bestClustAss)        
    #将被划分簇的一个中心，替换为划分后的两个中心
    centList[bestCentToSplit] = bestNewCents[0,:].tolist()[0] 
    centList.append(bestNewCents[1,:].tolist()[0])        
    #更新整体的聚类效果clusterAssment(类别，SSE)
    clusterAssment[nonzero(clusterAssment[:,0].A == bestCentToSplit)[0],:]= bestClustAss
    kMeans.datashow(dataSet,len(centList),mat(centList),clusterAssment)   
#return mat(centList), clusterAssment