Matt Haberland

Our messages crossed paths. I actually didn't see yours before I wrote mine. > No rush from my side, except not letting it stall unnecessarily. Yes, then I'd suggest making...

It did. You can see it in the [archive](https://mail.python.org/archives/list/[email protected]/).

@lorentzenchr do you mind if I push to your branch before this merges? There are some tests to be added (e.g. for `weights`), `dtype` still needs to be removed, and...

No problem @h-vetinari. I've approved, so I was just waiting to let you merge. I'd suggest committing the last [suggestion above](https://github.com/scipy/scipy/pull/17039#pullrequestreview-1162615968) with `[skip ci]` in the comment before merging. I...

No, I keep it simple. Like the title "ENH: stats: add `expectile` function". 'Nuff said, IMO.

I usually condense the stats release notes around release time. Just adding a separate line for now is fine.

Thanks @lorentzenchr @h-vetinari!

I think it's worth adding a note about unreliable results at points where the derivative doesn't exist. I'll do that today.

``` FAILED scipy/stats/tests/test_sampling.py::TestNumericalInverseHermite::test_basic_all_scipy_dists[anglit-shapes1] ``` Does not fail for me locally. @tirthasheshpatel @chrisb83 are you concerned about this? It's a RuntimeWarning. I doubt there's anything wrong, and warning when the algorithm...

@drew-parsons any luck here?