ocluster = function(datasam, classnum) {
    #有序样本聚类，输入datasam为样本数据阵，每一行为一个样本；
#输入classnum为要分的类数
#返回值result1为分类结果示意图
#各类的起始点存在变量breaks中
#输出三个矩阵 ra_dis:距离矩阵 leastlost:最小损失矩阵 classid:分类标识矩阵

#author:banmudi 2010.11



#样本数
    sam_n = dim(datasam)[1]

    #子函数，计算i-j个样本组成的类的半径
    radi = function(i, j) {
        #提取i-j个样本
        temp =as.matrix( datasam[i:j, ])
            mu = colMeans(matrix(temp,j-i+1))
            vec = apply(matrix(temp,j-i+1), 1, function(x) {
                x - mu
            })
            round(sum(apply(matrix(vec,j-i+1), 2, crossprod)),3)   

    }
   
   
    #计算距离矩阵
    ra_dis = matrix(0, sam_n, sam_n)
rownames(ra_dis) = 1:sam_n
    colnames(ra_dis) = 1:sam_n
    for (i in 1:(sam_n - 1)) {
        for (j in (i + 1):sam_n) {
            ra_dis[i, j] = radi(i, j)
            ra_dis[j, i] = radi(i, j)
        }
    }
   
    #最小损失矩阵，行为样本数，列为分类数
#leastlost[i,j]表示把1:i样本分成j类对应的最小损失
    leastlost = matrix(, sam_n - 1, sam_n - 1)
    rownames(leastlost) = 2:sam_n
    colnames(leastlost) = 2:sam_n
diag(leastlost) = 0
    #round(leastlost,3);

    #记录下对应的分类结点
    classid = matrix(, sam_n - 1, sam_n - 1)
    rownames(classid) = 2:sam_n
    colnames(classid) = 2:sam_n
    diag(classid) = 2:sam_n


    #分成两类时，填写最小损失阵的第一列
    leastlost[as.character(3:sam_n), "2"] = sapply(3:sam_n,
        function(xn) {
            min(ra_dis[1, 1:(xn - 1)] + ra_dis[2:xn, xn])
        })
    classid[as.character(3:sam_n), "2"] = sapply(3:sam_n, function(xn) {
        which((ra_dis[1, 1:(xn - 1)] + ra_dis[2:xn, xn]) == (min(ra_dis[1,
            1:(xn - 1)] + ra_dis[2:xn, xn])))[1] + 1
    })
    #分成j类时，填写最小损失阵的 第二列到最后一列
    for (j in as.character(3:(sam_n - 1))) {
        #分成j类
        leastlost[as.character((as.integer(j) + 1):sam_n), j] = sapply((as.integer(j) +
            1):sam_n, function(xn) {
            min(leastlost[as.character(j:xn - 1), as.character(as.integer(j) -
                1)] + ra_dis[j:xn, xn])
        })
       
        classid[as.character((as.integer(j) + 1):sam_n), j] = sapply((as.integer(j) +
            1):sam_n, function(xn) {
            a = which((leastlost[as.character(j:xn - 1), as.character(as.integer(j) -
                1)] + ra_dis[j:xn, xn]) == min(leastlost[as.character(j:xn -
                1), as.character(as.integer(j) - 1)] + ra_dis[j:xn,
                xn]))[1] + as.integer(j) - 1
        })
    }
   
    diag(classid) = 2:sam_n
   
    breaks = rep(0, 1, classnum)
    breaks[1] = 1
    breaks[classnum] = classid[as.character(sam_n), as.character(classnum)]
    flag = classnum - 1
    while (flag >= 2) {
        breaks[flag] = classid[as.character(breaks[flag + 1] -
            1), as.character(flag)]
        flag = flag - 1
    }

print("distance matrix:");#cat("\n")
print(ra_dis[2:sam_n,1:(sam_n-1)], na.print = ""); #输出距离矩阵
    print("leastlost matrix:")
print(leastlost[2:(sam_n-1),1:(sam_n-2)], na.print = ""); #输出最小损失矩阵
print("classid matrix:")
print(classid[2:(sam_n-1),1:(sam_n-2)], na.print = ""); #输出分类标识矩阵
    cat("\n")
print("result")
#画一个简单的分类示意图
    result1=NULL
    for (p in 1:sam_n) {
        result1 <- cat(result1,p, " ")
        for (w in 1:length(breaks)) {
            if (p == breaks[w] - 1) {
                result1 <- cat(result1, "||")
            }
        }
        if (p == sam_n)
           result1= cat(result1, "\n")
    }
}