Hi Maurizio, If your main interest is distribution modelling I would skip the GAM's and their inherent problems with overfitting, covariates with high correlations, etc. and use something like Random Forests or Generalized Boosting Models/Boosted Regression Trees. They have proven to be superior to GLM's and GAM's for species distribution modelling, but don't suffer from overfitting and can easily deal with large numbers of covariates, even when these are correlated. Both are described in dismo. For Random Forests you can also use BIOMOD.
Success, Henk Sierdsema Sovon Vogelonderzoek Nederland / Sovon Dutch Centre for Field Ornithology www.sovon.nl P.O. Box 6521 NL-6503 GA Nijmegen +31-247410445 The Netherlands ________________________________________ Van: r-sig-geo-boun...@r-project.org [r-sig-geo-boun...@r-project.org] namens Maurizio Marchi [mauriziomarch...@gmail.com] Verzonden: dinsdag 29 juli 2014 17:48 Aan: R-mailing list Onderwerp: [R-sig-Geo] GAM: which package? Hi everybody, I would like to try to work with generalized Additive Models to interpolate some climatic data and to build a Species Distribution Model. The aim is to check performances comparing them with works made by my colleague. I was wondering: Which R package is the most complete? here (http://cran.r-project.org/web/packages/dismo/vignettes/sdm.pdf) "mgcv" is suggested even if "gam" package seems the "pure" GAM. many thanks, -- Maurizio Marchi, Ph.D. student Florence, Italy ID skype: maurizioxyz Ubuntu 14.04 LTS linux user 552742 [[alternative HTML version deleted]] _______________________________________________ R-sig-Geo mailing list R-sig-Geo@r-project.org https://stat.ethz.ch/mailman/listinfo/r-sig-geo _______________________________________________ R-sig-Geo mailing list R-sig-Geo@r-project.org https://stat.ethz.ch/mailman/listinfo/r-sig-geo
