environmental variables

[Kerim ÇİÇEK]

Dear All,

In my model, I am using all worldclim variables. I am choosing "Non-spatial Partition" and jackknife. Then I am arranging all "feature classes" combination and 1-10 "regularization multipliers". In maxent model, some variables are reduced or omitted by Wallace. How can I see the effect of all variables for modeling? Thank you in advance.

Best, K.

[Gonzalo Pinilla-Buitrago]

Hi,

Maxent is selecting some of the variables to include in your model. If your variables are not used in the model, their contribution is 0. You can see the percentage of contribution with ENMeval::var.importance (not available in Wallace).

Merow, Cory, Matthew J. Smith, and John A. Silander Jr. "A practical guide to MaxEnt for modeling species’ distributions: what it does, and why inputs and settings matter." Ecography 36.10 (2013): 1058-1069. 

Best,

Gonzalo

[Jamie M. Kass]

I'll just add that if you want to see how species suitability responds to the variables kept in the model, check out the response curves in the Visualize component. Further, if you want all variables included, try a very low regularization value (even 0), but be warned that for feature classes like quadratic or hinge, low regularization will keep all the associated parameters (all quadratic and hinge coefficients) and make for a potentially very complex and maybe overfit model.

Jamie

--
You received this message because you are subscribed to the Google Groups "Wallace Ecological Modeling App" group.
To unsubscribe from this group and stop receiving emails from it, send an email to wallaceEcoMo...@googlegroups.com.
To post to this group, send email to wallac...@googlegroups.com.
Visit this group at https://groups.google.com/group/wallaceEcoMod.
To view this discussion on the web visit https://groups.google.com/d/msgid/wallaceEcoMod/893beced-0bd7-49a2-8bc3-f02ad2ac334c%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Patrick Kalonde]

Hello, I had a question regarding variable importance and I am just following up if currently we can check out variable contribution while/after using wallace. 

Thanks

Patrick

[Gonzalo Pinilla-Buitrago]

Hi Patrick,

Please check this question in our GitHub account. Link.

In summary, it is not possible to do it in Wallace. If you run your model using the session code and the maxent java option. You can access the "percent contribution" and "permutation importance" values in the @results slot of each model.

Best,

Gonzalo

[Danielle Del Castillo]

Hello,
I have done as suggested, running the models in R from the wallace session code, but my model object does not seem to have these data (percent contribution and permutation importance) in the @results slot. I have looked as well for these data in each of the @models slots.

Is there a code that I can use to extract or calculate these from the model object?

While I'm here, can someone also suggest a script to get variable response curves and model evaluation plots from the model object as well?

Regards,

Danielle