I solved my own problem. The Ensemble.Forecasting function needs to be edited to change MDA to FDA. Then it runs. <br><br><div class="gmail_quote">On Thu, Apr 15, 2010 at 10:59 AM, Alisa Wade <span dir="ltr"><<a href="mailto:alisaww@gmail.com">alisaww@gmail.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin: 0pt 0pt 0pt 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;">I am receiving an Ensemble.Forecasting error that I can not solve. Has anyone else seen this?<br>
<br>Here are my calls:<br><br>#Notes: <br>#MDA=T must be changed to FDA = T; incorrect in the manual, correct in the online help pages<br>
#Need this to make ANN to work (says invalid weight vector in nnet)<br>caseweights =matrix(ncol = ncol(spp.Mat), nrow = nrow(spp.Mat)); caseweights[,] =1 <br><br>Models(GLM = T, TypeGLM = "quad", Test = "AIC", GBM = T, No.trees = 3000, GAM = T, <br>
CTA = T, CV.tree = 50, ANN = T, CV.ann = 3, SRE = T, Perc025=T, Perc05=F, FDA = T, <br>MARS = T, RF = T, NbRunEval = 4, DataSplit = 80,Yweights=caseweights, <br>Roc=T, Optimized.Threshold.Roc=T, Kappa=T, TSS=T, KeepPredIndependent = F, VarImport=5, <br>
NbRepPA=0)<br><br>##all outputs look correct/exist; Evaluation.results.xxx all exist (except for SRE under Roc is NA)<br><br><br>Projection (Proj = env.Dat.Fut[,c(1, 3:20)], Proj.name = "NCAR_fut", GLM = T, <br>
GBM = T, GAM = T, CTA = T, ANN = T, SRE = T, Perc025 = T, Perc05 = F, <br>FDA = T, MARS = T, RF = T, BinRoc = T, BinKappa = T, BinTSS = T, <br>FiltRoc = T, FiltKappa = T, FiltTSS = T, repetition.models=TRUE)<br><br>##all outputs look correct/exist<br>
<br><br>And here is my ensemble call attempts and resulting errors:<br><br>##If I use FDA instead of MDA, it gives me an FDA error, no matter what weight.method or bin.method I use or no matter if FDA turned on or off<br>
<br>Ensemble.Forecasting(ANN = T, CTA=T, GAM = T, GBM = T, FDA=T, GLM = T, MARS = T, RF = T, SRE = F, Proj.name= "NCAR_fut", weight.method='Kappa', PCA.median=T, binary=T, bin.method='Kappa', Test=T, decay=2, repetition.models=F)<br>
<br>Error in Ensemble.Forecasting(ANN = T, CTA = T, GAM = T, GBM = T, FDA = T, : <br> unused argument(s) (FDA = T)<br clear="all"><br>##If I use MDA, then I get a parse error, no matter the bin method<br><br>Ensemble.Forecasting(ANN = T, CTA=T, GAM = T, GBM = T, MDA=T, GLM = T, MARS = T, RF = T, SRE = F, Proj.name= "NCAR_fut", weight.method='Roc', PCA.median=T, binary=T, bin.method='Roc', Test=T, decay=2, repetition.models=F)<br>
Error in !eval(parse(text = j)) : invalid argument type<br><br>Is this just a problem with ensemble.forecasting using the old MDA terminology when all the other functions have been changed to FDA?<br>Or am I doing something wrong?<br>
<br>Thanks for any ideas,<br>Alisa<br><font color="#888888">-- <br>*****************************************************<br>Alisa A. Wade<br>Postdoctoral Center Associate<br>National Center for Ecological Analysis and Synthesis<br>
<a href="mailto:wade@nceas.ucsb.edu" target="_blank">wade@nceas.ucsb.edu</a><br>
(406) 529-9722<br>home email: <a href="mailto:alisaww@gmail.com" target="_blank">alisaww@gmail.com</a><br>
</font></blockquote></div><br><br clear="all"><br>-- <br>*****************************************************<br>Alisa A. Wade<br>Postdoctoral Center Associate<br>National Center for Ecological Analysis and Synthesis<br>
<a href="mailto:wade@nceas.ucsb.edu">wade@nceas.ucsb.edu</a><br>(406) 529-9722<br>home email: <a href="mailto:alisaww@gmail.com">alisaww@gmail.com</a><br>