API: deprecate rename_axis / reindex_axis ?

[jorisvandenbossche]

Given we are adding the "labels, axis=0/1" idiom to reindex and rename itself (#12392, PR #17800), I think the specific reindex_axis and rename_axis will be superfluous.

Should be deprecate them?

[jreback]

yes

[TomAugspurger]

They aren't quite superfluous, as DataFrame.rename_axis can change .index.name and .columns.name:

In [13]: df = pd.DataFrame({"A": [1, 2, 3]})

In [14]: df.rename_axis("foo")
Out[14]:
     A
foo
0    1
1    2
2    3

In [15]: df.rename_axis("foo", axis=1)
Out[15]:
foo  A
0    1
1    2
2    3

I'm really regretting the approach I took in #11980 for overloading rename :/

[jorisvandenbossche]

For rename_axis that is true, and that is certainly useful. But we could deprecate the part that relabeling of the axis?

But I don't think it is the case for reindex_axis ?

I'm really regretting the approach I took in #11980 for overloading rename :/

Yes, but that is still something else no? s.name vs s.index.name ?

[TomAugspurger]

But we could deprecate the part that relabeling of the axis?

Ahh, yes, I suppose we could do that. I can start a PR now for it and reindex_axis.

[mpmX]

I use df = df.reindex_axis(sorted(df.columns), axis=1) to sort columns alphabetically and got a FutureWarning once in the past.

FutureWarning: '.reindex_axis' is deprecated and will be removed in a future version. Use '.reindex' instead.

The warning did not occur since then, so I wonder what the current state is.

[toobaz]

I use df = df.reindex_axis(sorted(df.columns), axis=1) to sort columns alphabetically and got a FutureWarning once in the past.

Got the same problem in old code... solved with df = df.loc[:, sorted(df.columns)]