Jérémie du Boisberranger

out of curiosity, what is the parameter that you want to set to such a big int ?

CI is failing because this PR introduces many FutureWarnings. You need to check every place where we call `(MiniBatch)NMF` or `non_negative_factorization` where `n_components` is left to its default value and...

Thinking about it again I think we should explicitly ignore the warnings instead of setting `n_components="auto"` in the tests. Sorry for my previous misleading comment. This can be done by...

> Should we also add n_components="auto" this example? I haven't found other places where the warning could occur Yes, we don't want warnings in the examples either

Thanks for the PR @aanaseer. As discussed in https://github.com/scikit-learn/scikit-learn/pull/26542, this is a kind of function that happens to be public but should really be part of a "dev API", in...

Thanks for the PR @rand0wn > I rather not add validation to functions in this file. I'm not sure for all functions in this file. This one for sure should...

Thanks for the PR @StefanieSenger. With @glemaitre we agree that this is not the kind of function that we want to validate. It's typically the kind of function that is...

> Import speed (point 2): what kind of applications can we think of where this is a limiting/annoying factor? From my personal POV I don't have things that bother me...

Why do we need to vendor `lazy_loader` ? can't we make it a dependency ?

> At least I read Jeremie's comment more as ideal speculation over a drink, than the result of measuring the speed up for a typical user. That's exactly what it...