Maybe look at bringing in tidymodels for ensemble modeling

[danlwarren]

Nice little tutorial here:

http://amunategui.github.io/blending-models/

It seems like we could bring in caret in such a way that you pass in a list of model types (with a default list, presumably). This would likely require us to write a predict function for caret models onto raster models, but that shouldn't be too hard. Depending on the predict syntax for caret, we might even be able to use one of the existing raster::predict functions by tricking it with a temp copy of the caret model with the class values reset.

Another nice thing about doing this is that it wouldn't necessarily just be for ensembles; we could use this to open up a whole bunch of other methods that caret is a wrapper for, just by passing a single model type in the list.

[danlwarren]

Actually we should really do this in the context of tidymodels, not caret. I think we might actually be able to create a central function (enmtools.tidymodel) and then just convert all of the enmtools.model functions to call it with the requisite arguments. There's a LOT of cool stuff to be done here, and it could make ENMTools simultaneously easier to maintain AND more flexible for users.

https://www.youtube.com/watch?v=z57i2GVcdww

[danlwarren]

Russell found this, which seems potentially useful:

https://www.tidymodels.org/learn/

[danlwarren]

Even better, here's a whole book!

https://www.tmwr.org/

[rdinnager]

Yes! I just saw this pop up on my twitter feed as well. Got some reading to do!