R语言实现K-Means算法数据集iris

博主代码根据机器挖掘实战编写的.那本书用的是Python.Step By Step. R语言有函数包实现

也可以调用，里面的四个algorithm我怎么看着都差不多，所以我决定写一下，明天仔细看一下

kmeans(x, centers, iter.max = 10, nstart = 1,

algorithm = c(“Hartigan-Wong”, “Lloyd”, “Forgy”,

“MacQueen”))

代码
#加载iris函数
loadData <- function(){
data(iris)
#要把Species列去掉
dataSet =  iris[,-ncol(iris)]
return (dataSet)
}
#计算欧几里得距离
distEnclud <- function(vecA,vecB){
distEnclud = sqrt(apply((vecA-vecB)^2,1,sum))
return (distEnclud)
}
#设置初始族
randCent <- function(dataSet,k){
n = ncol(dataSet)
#建立一个空矩阵
centroids = matrix(data = NA, nrow = k, ncol = n, byrow = FALSE,dimnames = NULL)
#随机设置初始族
for(j in 1:n){
minJ = min(dataSet[,j])
rangeJ = max(dataSet[,j]) - minJ
#随机生成每一列的初始簇
centroids[,j] = minJ + rangeJ *runif(k)
}
return (centroids)
}
kMeans <- function(dataSet,k,distEnclud, randCent){
m = nrow(dataSet)
#记录矩阵，一列记录簇索引值，即类别，第二列记录误差
clusterAssment = matrix(data = 0,  nrow = m , ncol = 2,byrow = FALSE,dimnames = NULL)
#初始族
centroids = randCent(dataSet,k)
#初始簇列名和数据集列名相同
colnames(centroids) = colnames(dataSet)

clusterChanged = TRUE
#设置结束变量，如果为TRUE 说明族中心点和分类结果还有变化，所以要继续分类，若没有改变则设置为FALSE 跳出循环
while (clusterChanged) {
clusterChanged = FALSE
for(i in 1: m){
minDist = Inf
minIndex = -1
for(j in 1:k){
distJI = distEnclud(dataSet[i,],centroids[j,])
if(is.na(distJI)) distJI = Inf
if(distJI < minDist){
minDist = distJI
minIndex = j
}
}
if(clusterAssment[i,1] != minIndex){
clusterChanged = TRUE
clusterAssment[i,1] = minIndex
}
clusterAssment[i,2] = minDist^2
}

#每计算一遍 都要 updata center
for(cent in 1:k){
ptsCluster = dataSet[which(clusterAssment[,1] == cent),]
centroids[cent,] = apply(ptsCluster,2,mean)
}
}
#小技巧 R语言返回几个矩阵时 可用list将其组合，返回后再如取list元素一边取出就好
out = list(clusterAssment = clusterAssment ,centroids = centroids)
return (out)
}
###########调用函数代码，只需3行，简单吧
#dataSet = loadData()
#head(dataSet)
#output = kMeans(dataSet,2,distEnclud,randCent)

#绘制Sepal.Length，Sepal.Width 两列
plot(dataSet[c("Sepal.Length","Sepal.Width")],col = output$clusterAssment[,1])
points(output$centroids[,1],output$centroids[,2],col = 1:2,pch = 8,cex = 2)
#绘制Petal.Length，Petal.Width 两列
plot(dataSet[c("Petal.Length","Petal.Width")],col = output$clusterAssment[,1])
points(output$centroids[,3],output$centroids[,4],col = 1:2,pch = 8,cex = 2)
table(iris$Species, output$clusterAssment[,1])

根据Sepal.Length，Sepal.Width 两列两列画出的图形



根据Petal.Length，Petal.Width 两列画出的图形