[Biomod-commits] qual.th and the models in the ensemble

[Wilfried Thuiller]

Could you please send the suite of calls you made please? 

In any case, I will not put a so high criteria give your results. A quality threshold for 0.4 or 0.5 will be sufficient to get rid off of the crappy models.



Le 20 déc. 2011 à 15:19, Frederico Mestre a écrit :

> Hello:
> 
> I'm not quite sure how to select the criteria, but if it is the same as the  bin.method, then I'm using TSS.
> 
> I've checked these tables "Evaluation.results.TSS" (bellow) and, the cross validation value is rarely above 0.7. 
> But even those few models that have a cross validation >0.7 should be part of an esemble with qual.th=0.7, right?
> 
> Anyway, I feel I'm missing something.
> 
> Thanks,
> 
> Frederico
> 
> mc_PA2_rep4
>      Cross.validation indepdt.data total.score Cutoff Sensitivity Specificity
> ANN              0.569         none   0.6196809 186.03    81.11702    80.85106
> CTA               0.619         none   0.7723404 301.94    89.09574    88.13830
> GAM             0.588         none   0.5957447 504.35    88.82979    70.74468
> GBM              0.704         none   0.7781915 520.84    91.48936    86.32979
> GLM              0.529         none   0.5074468 462.95    86.70213    64.04255
> MARS           0.501         none   0.5664894 112.13    85.10638    71.54255
> FDA              0.603         none   0.5978723 131.59    83.77660    76.01064
> RF                0.702         none   0.8781915 244.25    92.55319    95.26596
> SRE             0.190         none   0.2617021  10.00    62.23404    63.93617
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 2011/12/20 Wilfried Thuiller <wilfried.thuiller at ujf-grenoble.fr>
> Hi there,
> 
> Have you checked the evaluation tables? Which criteria do you use for the quality threshold? 
> 
> Best
> Wilfried
> 
> 
> Le 20 déc. 2011 à 14:58, Frederico Mestre a écrit :
> 
>> Hello all:
>> 
>> 
>> 
>> I’m running some ensembles and projecting them in a GIS.
>> 
>> 
>> 
>> First, I tried to run an ensemble using all the models (not setting any value to “qual.th”). And then I visualize the results in a GIS. The result maps looked pretty good (apparently the models correctly classified areas where the species occurs as those where it is more probable, and it might have predicted the locations of some new populations).
>> 
>> 
>> 
>> These results, apparently, are good. But, when I run another ensemble trying to use only the best models (setting the qual.th=0.8, or even 0.7), I get the error:
>> 
>> 
>> 
>> Error in if (sum(ProbData) != 0) { :
>> 
>> missing value where TRUE/FALSE needed
>> 
>> 
>> 
>> What does this mean? None of the models has a value higher than 0.8 (or 0.7)? How is it possible if the models apparently are good in classifying the areas where the species occurs?
>> 
>> What am I missing?
>> 
>> 
>> 
>> Thanks,
>> 
>> Frederico
>> 
>> _______________________________________________
>> 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
> 
> --------------------------
> Dr. Wilfried Thuiller
> Laboratoire d'Ecologie Alpine, UMR CNRS 5553
> Université Joseph Fourier
> BP53, 38041 Grenoble cedex 9, France
> tel: +33 (0)4 76 51 44 97
> fax: +33 (0)4 76 51 42 79
> 
> Email: wilfried.thuiller at ujf-grenoble.fr
> Personal website: http://www.will.chez-alice.fr
> Team website: http://www-leca.ujf-grenoble.fr/equipes/emabio.htm
> 
> FP6 European MACIS project: http://www.macis-project.net
> FP6 European EcoChange project: http://www.ecochange-project.eu
> 
> 
> 
> 
> 
> 
> 

--------------------------
Dr. Wilfried Thuiller
Laboratoire d'Ecologie Alpine, UMR CNRS 5553
Université Joseph Fourier
BP53, 38041 Grenoble cedex 9, France
tel: +33 (0)4 76 51 44 97
fax: +33 (0)4 76 51 42 79

Email: wilfried.thuiller at ujf-grenoble.fr
Personal website: http://www.will.chez-alice.fr
Team website: http://www-leca.ujf-grenoble.fr/equipes/emabio.htm

FP6 European MACIS project: http://www.macis-project.net
FP6 European EcoChange project: http://www.ecochange-project.eu






-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/biomod-commits/attachments/20111220/6ab9f5e9/attachment-0001.htm>