implementation of fold.vector parameter in gbm.auto ?

[clescoat]

Hi,

I have zero-inflated data which lead me to your package for BRT analysis. However, there is also spatial autocorrelation in my data. I currently use the fold.vector parameter in gbm.step (dismo) to perform spatial cross-validation with blocks created by BlockCV (Valavi et a., 2018). It seems that it's not available in gbm.auto, is it something which could be implemented?

Thanks for your help !

[SimonDedman]

Howdy, sorry for the slow reply, been a crazy few weeks.

This is the first I've heard about fold.vector. I've forked (kinda) gbm.step within gbm.auto, and gbm.auto includes a ... parameter which allows any parameter to be passed to it.

So: it might just work already! Give it a try & LMK how you get on.

How do you actually use this? Like much of dismo, there's no documentation on this. BlockCV link.

Cheers!