Christopher Anderson

uses of the reference data https://www.tandfonline.com/doi/abs/10.1080/2150704X.2018.1465611 https://www.sciencedirect.com/science/article/abs/pii/S0034425715000395 check this out, too? https://www.sciencedirect.com/science/article/abs/pii/S0034425716300475

To address the difference in spectral fidelity, we should also use just the overlapping spectral bands between sources (that is, use the 6 most comparable S2 bands to L8).

@PC-FSU let me know if you plan to finish up this contribution. If not, I'll go ahead and merge it and make the remaining updates. Cheers,

I've reviewed this PR again and want to document the next steps. PF-FSU has contributed a large PR with several pieces. There are still a number of changes I want...

Yes @bweeding there is a [model.permutation_importance_plot()](https://earth-chris.github.io/elapid/module/models/#elapid.models.SDMMixin.permutation_importance_plot) function and a [model.permutation_importance_scores()](https://earth-chris.github.io/elapid/module/models/#elapid.models.SDMMixin.permutation_importance_scores) function. The documentation points to an `sklearn` article on how to interpret such scores, and how to be critical of...