Starting from May 2009, I have switched to R for scientific computing. It seems more convenient to me. And, geeky reason, I can use VIm as my IDE.

R scripts for feature selection

• 2009sgai-featuresel.R, forward feature selection, using regression tree and support vector regression as predictive models, described in this post

Update 2011-11-21: There’s a lot more scripts which I’ve used in my PhD thesis. The spatial variable importance framework is something which I couldn’t find in an R package, so I developed it. Same for the spatial clustering for management zone delineation. It might be much quicker in a C(++) library, but it runs in R natively. Currently the scripts are available on demand only, since they’re not yet in a fully publishable (clean) form. Anyway, R is definitely my language of choice for the tasks I’m handling. Love it.