[Biomod-commits] Problem and questions

[Damien Georges]

|*Dear,

First, I advise you to get the last version of biomod2 available on 
R-Forge (2.1.15) :

install.packages("biomod2", repos="http://R-Forge.R-project.org")

Many little bugs have been fixed and the one you mentioned should be one 
of them.


Moreover, in last versions it is possible to user to defined a 
"home-made" set of pseudo-absences.. PA.strategy='user.defined', 
PA.table= .... in BIOMOD_FormatinData() function. Might be usefull in 
your case.



Q1.

No user friendly function is implemented yet.. waiting for this, you 
could acces this info going into your "BIOMOD.EnsembleModeling.out" object..


Weights used for probability weighted mean are stored in the slot 
@em.weight (i.e. myBiomodEM at em.weight) for all built ensemble models. 
This weights are applied to models stored into
@em.res$---The_ensemble_model_name---$em.models.kept. It's a bit 
troublesome so I will create a little fuction to do it for you.
Here an example :

myBiomodEM <- BIOMOD_EnsembleModeling(...)

# the PWM weights
myBiomodEM at em.weight
$GuloGulo_TotalConsensus_EMbyTSS
[1] 0.3086854 0.3376369 0.3536776

# the correspunding models
myBiomodEM at em.res$GuloGulo_TotalConsensus_EMbyTSS$em.models.kept
[1] "GuloGulo_AllData_RUN1_SRE" "GuloGulo_AllData_RUN1_CTA" 
"GuloGulo_AllData_RUN1_RF"


Q2.
To export projection you will be able to laod in your GIS software... In 
BIOMOD_Projection() function, set the arg do.stack=FALSE
(for export projections layer by layer) and output.format='.img' (to 
export raster into .img format.. The only supported yet)

Q3.
Not directly but you can load the .grd produced with stack() function 
and then export layers with writeRaster function.. Note that you can 
follow the same procedure if you want to export your projections into 
another format than .img.

Hope that helps,

Best,

Damien.
*|**|*
*|**


On 15/03/2013 10:45, J DL wrote:
> Dear all,
>
> I had  a problem that i manage to solve although outside biomod, and some
> questions
> 1.1 ) Problem
> I am running Biomod package 2.0.3.
> I got the following error.
> It reads: !Error in abind(calib.lines, mod.prep.dat[[pa]]$calibLines, along
> = 3) : arg 'X2' has dims=367, 2, 1; but need dims=370, 2, X
>
> After many checkings I discovered that the problem came out from the pseudo
> absences generation. I managed to solve it by creating them manually in gis.
>  From my understanding, its looks like that the point are created outside
> the raster dimension.
> it is this solve in biomod version 2.1.7?
>
> 2) Questions.
>
> 2.1) Is there any fucntion to display in the screen the weight that are
> given to each model in the probabilities mean decay ensemble model?
>
> 2.2) Is It possible to export a single model projection to raster, tiff,
> grid or ascii, in order to charge it in any gis software.
> 2.3) Is it possible to export a singel ensamble model projection to raster,
> tiff, grid or ascii, in order to charge it in any gis software.
>
> Thanks
>
> Joaquin
> _______________________________________________
> 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