search

shandongfx / nimbios_enm

NIMBIOS ENM
4 stars 1 forks source link

Problem with definition of categorical variables #1

Open IlhemB opened 4 years ago

IlhemB commented 4 years ago

Hello, I have categorical variables that are declared as “character” in the dataframe “pder” but even though in the results of maxent it says “Environmental layers used (all continuous)”. I read in the documentation that the argument “factors ” is used only when the environmental variables are defined as raster file. But how can we define the categorical variable when we work with data.frame?

shandongfx commented 4 years ago

If you could provide a script with demo data, I may be able to help debug it.

A quick though is, try to do another run with replacing that column with more obvious characters, such as "fact_1", "fact_2", "fact_3".

IlhemB commented 4 years ago

Hi, Thank you very much for your reply. Actually I was trying to figure out the problem, so I noticed that if we work with "rasters" rather than "data.frame" we don't have this problem for example: mod <- dismo::maxent(x=env, ## env conditions (raster) p=Occ_final, ## Occ_final a=bg, method = "randomkfold", kfolds=10, args = c( 'outputformat=cloglog', 'autofeature=true', 'betamultiplier=4', 'responsecurves=true', 'jackknife=true', 'writeplotdata=true', 'askoverwrite=false', 'writebackgroundpredictions=true', 'writemess=false'), factors=c("fact_1", "fact_2", "fact_3") )

I’m using also “'randomkfold'” method to partition the occurence and the background points. I want to extract the outputs (ROC, prediction, response curves…) as the average of the replications. Is there any way to do it ?

Thank you in advance Best