[R] bivariate interpolation

[Adelchi Azzalini]

On Mon, 2 Apr 2007 22:26:06 -0400, eric lee wrote:

EL> Hi.  I'm trying to take a data set with two independent and one dependent
EL> variable and enter a x,y value to predict the dependent with a
EL> nonparametric technique.  I've been using interpp in the akima package,
EL> (windows xp, R 2.4.1), but get values that are orders of magnitude off
EL> when the predictors are slightly out of the range of the data set.  Can
EL> you recommend a function for me?  I've read that predict.loess has the
EL> same problem or just lists NA.  Smooth.spline looks good, but can only do
EL> one predictor instead of bivariate.  I've tried help.search("predict"),
EL> but couldn't find what I needed.  Thanks.
EL> 

one option is to use sm.regression of package sm, along these lines

 x <- cbind(runif(100,-2, 2), runif(100,-2, 2))
 y <- x[,1]^2 + x[,2]^2 + 0.7*x[,1]*x[,2] + rnorm(100)/3
 sm.regression(x,y)
 ev.pt <- rbind(c(-2,-2),c(2,2), c(2,-2),c(-2,2))
 a <- sm.regression(x,y, eval.points=ev.pt,  eval.grid=FALSE)
-- 
Adelchi Azzalini  <azzalini a stat.unipd.it>
Dipart.Scienze Statistiche, Università di Padova, Italia
tel. +39 049 8274147,  http://azzalini.stat.unipd.it/