David Watson

For a little background: Breiman & Cutler introduced permutation importance in their [random forest classification manual](https://www.stat.berkeley.edu/~breiman/RandomForests/cc_home.htm#varimp). Strobl & Zeileis wrote an [interesting article](https://eeecon.uibk.ac.at/~zeileis/papers/Strobl+Zeileis-2008.pdf) a few years later advocating the unscaled...

I see. So this would apply in cases where the `lm` function itself encodes those transformations, correct? Like `lm(y ~ I(x^2) + x*z)` or something? Perhaps there could be an...

Cool! Did not know about the `type = "terms"` option. It looks like that's just exactly what we want, right? The function appears to do mean centering by default (see...

Sounds good! I thought your original concern was that, while the independence assumption is probably false in many real world cases, it is self-evidently bogus when one variable is a...

That makes sense, thanks! Resolves my issue. Perhaps this could be a bit clearer in the documenation? Or maybe instead of an error, `eigs_sym()` could just run `eigs()` internally when...