Christian Lorentzen

>There are some untested lines which it seems they're legit and should be tested Except for about one occasion, the lines not tested according to codecov are just hard to...

#27639 is merged. So we could resume to this PR. As to the tests, we should have almost the same as in sklearn/ensemble/_hist_gradient_boosting/tests/test_monotonic_contraints.py.

> There were discussions around whether line search should be performed when using monotonic constraints AFAIU, line search must be performed with the addition that node values (after ls) trespassing...

> While I did sketch out a conceptual fix, If you have a fix for the recursive tree method, even if very rough, we'd be interested! The fix is the...

@mayer79 Do you have any insights or recommendations here?

> I think it would be good to mention this in the docu We should definitely document this. PR very welcome.

@mayer79 Thanks for working on this important inspection tool. To get rid of the linter issues, you might use a pre-commit hook, see https://scikit-learn.org/dev/developers/contributing.html#how-to-contribute. @amueller @glemaitre @adrinjalali ping as this...

> I keep struggling over the fact that "Friedman's H-statistic" is actually an H-squared. The naming will pop up during further review anyway. One possibility would be `h2_statistics`.

> Could you add a test for a known (linear) predict function, where the H-statistic is available analytically? (Like a mocking-Estimator that does not need fitting). What about $m(x_1, x_2)...

> No idea how to fix the failing checks... might it be due to some different pandas on Debian? You can see in the [CI](https://dev.azure.com/scikit-learn/scikit-learn/_build/results?buildId=66118&view=logs&j=aabdcdc3-bb64-5414-b357-ed024fe8659e&t=b7b3ba55-d585-563b-a032-f235636c22b0&l=4911) that Linux_Docker debian_atlas_32bit installed pandas...