Michael Mayer

Any progress on this issue? I would love to see it fixed.

Oops, wrong button ;-). Sounds reasonable! My suggestion only makes a difference if a variable list is passed as well. If there is no such list, it makes sense to...

The current implementation is how you would naturally define interaction constraints: Each leaf node with see only variables from one variable group allowed to interact. This generates a boosted trees...

@madprogramer : Thanks for pushing this! I'd love to see these geoms available. The current release of "shapviz" at least plays well together with plotly regarding the dependence plot and...

Just a comment: In R base, the function `max.col` seems to be very efficient already. It does a "rowWhichMaxs"

@samronsin @jjerphan may I ask about the next steps here? If I correctly followed the issues, this would be the basis to have random forests with monotonic constraints. https://github.com/scikit-learn/scikit-learn/issues/18982

> Actually @mayer79 this PR already implements monotonicity constraints for trees ensembles (including Random Forests), so this would be it, no next step. > > I'll address the last comments...

@hannes, I would love to see this part being implemented, but based on @lorentzenchr's comment, it seems out of scope for an "ousider". So your help would be highly appreciated...

Thanks, @ak2k . This is indeed a good idea. If the arrow table does not fit into RAM: Is it possible to do this in a chunkwise fashion? I could...

It is a good question. I share your concerns and would love to see articles on that topic. SHAP is as good or bad as the trained model. I.e. tuning...