CLN: named parameters for GroupBy.(mean|median|var|std)

[topper-123]

Drops *args & **kwargs, replace with named parameters for groupby methods mean, median, var & std. Similar to #31473.

This PR has the side effect that the raised error when a parameter is not allowed, is now TypeError instead of UnsupportedFunctionCall, so technically a API change...

[WillAyd]

looks good to me. I'd be ok with 1.1 but maybe worth a 2.0 target as well

[topper-123]

Maybe a decorator similar to deprecate_kwargs could be added. That way we'd get the correct signature, and UnsupportedFunctionCall would be raised in the decorator.

[jreback]

looks good to me. I'd be ok with 1.1 but maybe worth a 2.0 target as well

we don’t have the bandwidth or infrastructure to handle this type of versioning
if it’s not targeted then would rather close

this is fine for 1.1

[topper-123]

Ok, so I don't do a decorator solution?

[jreback]

lgtm. a question.

[jreback]

can you update doc-strings for these parameters (or are they already ok with the Appenders?)

[topper-123]

Ok, I've added doc strings.

Also, I've found out that Groupby.median actually didn't accept min_count in master, but raised an AssertionError. So I removed the parameter from that method signature.

[jreback]

thanks @topper-123

[jreback]

Ok, I've added doc strings.

Also, I've found out that Groupby.median actually didn't accept min_count in master, but raised an AssertionError. So I removed the parameter from that method signature.

happy to have an issue / PR about this.

[topper-123]

I did fix that is this PR, so no issue necesary.

So in short, in v1.0.0/master we'd have that grp.mean(min_count=1) raised a AssertionError, while now it raised a TypeError, which is the same error code as we'd get for df.mean(min_count=1), so is the correct error type, IMO.