[Biomod-commits] variables problem

[Damien Georges]

Dear Maria,

Are you sure that suitable variables combination in present occurs in 
past conditions? (if not, that can explain the low probabilities you got)

I guess you did your variable selection on current environmental data. 
So you removed the variables that were too correlated one to another. 
Did you check current correlated variables are the same than the past 
ones? Maybe you have made a too drastic variables selection and remove 
some useful variables to explain past distribution.

Tell me if that helps,

Best,

Damien.

On 05/11/2012 12:34, Maria Guerrina wrote:
> Dear all,
> I removed variables highlightly correlated and I selected 8 of them.
> Past models made using 8 variables are strongly different by models made using
> all variables. Past models made using 8 variables don't predict habitat
> suitability for the species (values are very low).
> In the present there isn't any difference.
> Can someone explain to me why?
>
> Maria Guerrina
> PhD Student
> Università di Genova
> DISTAV
> Corso Dogali 1M
> I - 16136 GENOVA (Italy)
> _______________________________________________
> Biomod-commits mailing list
> Biomod-commits at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/biomod-commits