lab <- liste

predicted <- kmeans(lab[,c(2,3,4)],10,nstart = 20)

tahmin <- cbind.data.frame(lab, predicted$cluster)

colnames(tahmin) <- c("isim", "lab1", "lab2", "lab3", "Data", "Class")


View(tahmin, "lab")

## dbscan
 library("dbscan")
 cl <- hdbscan(U, minPts = 5)
 
#plot outlier scores

 plot(sort(cl$outlier_scores), ylab="hdbscan score")
abline(h=quantile(cl$outlier_scores,c(.9,.95)), col=c("blue","red"))
legend("bottomright", legend = c("90 percentile", "95 Percentile"), lty=1, col=c("blue","red") , bty = "n")

#find which rows are outlier
which(cl$outlier_scores >quantile(cl$outlier_scores,0.95))