Hi all, I'm looking for a way to use loess model (?loess) with a different percentile (instead of median). I hope I made myself clear enough. Here's and example: a<-data.frame(x=seq(1:200),y=200*rnorm(200)) lss<-loess(y~x,a) predict(lss, 120) will give a local median of that dataset. How about a local 99 percentile? Thanks.- -- Regards, Anh Tran [[alternative HTML version deleted]]