library(MASS) # kk<-kmeans(iris[,1:4],3) table(kk$cluster,iris$Species) # ll<-lda(iris[,1:4],iris$Species) pl<-predict(ll,iris[,1:4]) table(pl$class,iris$Species) ######################################### W<-rep(0,10) for (k in 2:10) { kk<-kmeans(iris[,1:4],k) W[k]<-kk$tot.with } mu<-apply(iris[,1:4],2,mean) iriss<-iris[,1:4]-t(matrix(rep(mu,dim(iris)[1]),4,dim(iris)[1])) W[1]<-sum(iriss^2) plot(seq(1,10),W,xlab="K",ylab="Within-SS",type="l") ############# kk<-kmeans(iris[,1:4],3) plot(iris[,2:3],col=as.numeric(iris$Species)+1,pch=kk$cluster) # kk<-kmeans(iris[,1:4],2) plot(iris[,2:3],col=as.numeric(iris$Species)+1,pch=kk$cluster) #########################333 ###### pp<-pam(iris[,1:4],3) plot(pp) # plot(iris[,2:3],col=as.numeric(iris$Species)+1,pch=pp$cluster) # W<-rep(0,10) for (k in 2:10) { kk<-pam(dist(iris[,1:4]),k) W[k]<-kk$sil$avg } mu<-apply(iris[,1:4],2,mean) iriss<-iris[,1:4]-t(matrix(rep(mu,dim(iris)[1]),4,dim(iris)[1])) plot(seq(2,10),W[2:10],xlab="K",ylab="Silhouette Width",type="l") ###################### hh<-hclust(dist(iris[,1:4])) plot(hh,label=iris$Species) ### hh<-hclust(dist(iris[,1:4]),"single") plot(hh,label=iris$Species) ### cc<-cor(t(iris[,1:4])) hh<-hclust(as.dist(1-cc)) plot(hh,label=iris$Species) #