class Kohonen(object):def __init__(self):self.lratemax = 0.8        #最大学习率--欧式距离self.lratemin = 0.05       #最小学习率--欧式距离self.rmax     = 5.0        #最大聚类半径--根据数据集self.rmin     = 0.5        #最小聚类半径--根据数据集self.Steps    = 1000       #迭代次数self.lratelist = []        #学习率收敛曲线self.rlist     = []        #学习半径收敛曲线self.w         = []        #权重向量组self.M         = 2         #M*N聚类的总数self.N         = 2         #M/N表示邻域的参数self.dataMat   = []        #外部导入数据集self.classLabel = []       #聚类后的类别标签

def normlize(self,dataMat):         #数据标准化归一化[m,n] = shape(dataMat)for i in xrange(n-1):dataMat[:,i] = (dataMat[:,i]-mean(dataMat[:,i]))/(std(dataMat[:,i])+1.0e-10)return dataMat

    def distEclud(vecA,vecB):#欧式距离eps = 1.0e-6return linalg.norm(vecA-vecB) + eps

 def loadDataSet(self,filename):     #加载数据集numFeat = len(open(filename).readline().split('\t'))-1fr      = open(filename)for line in fr.readlines():lineArr = []curLine = line.strip().split('\t')lineArr.append(float(curLine[0]))lineArr.append(float(curLine[1]))self.dataMat.append(lineArr)self.dataMat = mat(self.dataMat)

#coding:utf-8
from numpy import *class Kohonen(object):def __init__(self):self.lratemax = 0.8        #最大学习率--欧式距离self.lratemin = 0.05       #最小学习率--欧式距离self.rmax     = 5.0        #最大聚类半径--根据数据集self.rmin     = 0.5        #最小聚类半径--根据数据集self.Steps    = 1000       #迭代次数self.lratelist = []        #学习率收敛曲线self.rlist     = []        #学习半径收敛曲线self.w         = []        #权重向量组self.M         = 2         #M*N聚类的总数self.N         = 2         #M/N表示邻域的参数self.dataMat   = []        #外部导入数据集self.classLabel = []       #聚类后的类别标签def ratecalc(self,i):               #学习率和半径Learn_rate = self.lratemax-((i+1.0)*(self.lratemax-self.lratemin))/self.StepsR_rate     = self.rmax-((i+1.0)*(self.rmax-self.rmin))/self.Stepsreturn  Learn_rate,R_ratedef normlize(self,dataMat):         #数据标准化归一化[m,n] = shape(dataMat)for i in xrange(n-1):dataMat[:,i] = (dataMat[:,i]-mean(dataMat[:,i]))/(std(dataMat[:,i])+1.0e-10)return dataMatdef distEclud(self,vecA,vecB):#欧式距离eps = 1.0e-6data = vecA-vecBdata = linalg.norm(vecA-vecB) + epsreturn linalg.norm(vecA-vecB) + epsdef loadDataSet(self,filename):     #加载数据集numFeat = len(open(filename).readline().split('\t'))-1fr      = open(filename)for line in fr.readlines():lineArr = []curLine = line.strip().split(' ')lineArr.append(float(curLine[0]))lineArr.append(float(curLine[1]))self.dataMat.append(lineArr)self.dataMat = mat(self.dataMat)def init_grid(self):k = 0 #构建第二层网络模型grid = mat(zeros((self.M*self.N,2)))for i in xrange(self.M):for j in xrange(self.N):grid[k,:] = [i,j]k += 1return griddef train(self):dm,dn = shape(self.dataMat)                 #1.构建输入网络normDataset = self.normlize(self.dataMat)   #归一化数据grid        = self.init_grid()              #初始化第二层分类器self.w      = random.rand(self.M*self.N,dn)#3.随机初始化两层之间的权重值# distM       = self.distEclud           #确定距离公式#4.迭代求解if self.Steps < 5*dm:self.Steps = 5*dm                        #设定最小迭代次数for i in xrange(self.Steps):lrate,r = self.ratecalc(i)               #1)计算当前迭代次数下的学习率和分类半径
            self.lratelist.append(lrate)self.rlist.append(r)#2)随机生成样本索引，并抽取一个样本k = random.randint(0,dm)mySample = normDataset[k,:]#3)计算最优节点：返回最小距离的索引值minIndx =[self.distEclud (mySample,i) for i in self.w]minIndx = minIndx.index(min(minIndx))# self.w[minIndx,:] = self.w[minIndx,:]+lrate*(mySample[0]-self.w[minIndx,:])#4)计算邻域d1 = ceil(minIndx/self.M)  #计算此节点在第二层矩阵中的位置d2 = mod(minIndx,self.M)distMat = [self.distEclud(mat([d1,d2]),i) for i in grid]nodelindx = (array(distMat) < r).nonzero()for j in xrange (shape(self.w)[1]):  #5)按列更新权重if sum(nodelindx == j):self.w[:,j] = self.w[:,j]+lrate*(mySample[0]-self.w[:,j])#主循环结束self.classLabel = range(dm) #分配和存储聚类后的类别标签for i in xrange(dm):self.classLabel[i] = distM(normDataset[i,:],self.w).argmin()self.classLabel = mat(self.classLabel)SMOnet = Kohonen()
SMOnet.loadDataSet('testSet2.txt')
SMOnet.train()