library(mlbench)
library(mclust)

data(PimaIndiansDiabetes)
help(PimaIndiansDiabetes)

head(PimaIndiansDiabetes)
X <- PimaIndiansDiabetes[,1:8]

mclust.obj <- Mclust(scale(X))
plot(mclust.obj)
print(mclust.obj)

I <- mclust.obj$uncertainty < 0.25
mean(I)

clusters <- unique(mclust.obj$classification)
M <- matrix(0,length(clusters),10)
for (i in 1:length(clusters)){

	I <- mclust.obj$classification==clusters[i]
	M[i,1:8] <- colMeans(X[I,])
	J <- PimaIndiansDiabetes[I,9]== "pos"
	M[i,9] <- mean(J)
	M[i,10] <- sum(I)
}

colnames(M) <- c(colnames(PimaIndiansDiabetes)[1:8],"Prop. yes","N")
rownames(M) <- clusters
options(digits=3)
M


library(lattice)
mds.obj <- cmdscale(dist(scale(X)),k=2) 
xyplot(mds.obj[,2]~mds.obj[,1],groups=mclust.obj$classification,pch=1:16)