REGR: Performance of DataFrame reduction ops with axis=1

[rhshadrach]

• closes #xxxx (Replace xxxx with the GitHub issue number)
• Tests added and passed if fixing a bug or adding a new feature
• All code checks passed.
• Added type annotations to new arguments/methods/functions.
• Added an entry in the latest doc/source/whatsnew/vX.X.X.rst file if fixing a bug or adding a new feature.

Siphoning off the largest impact from #52083.

shape = 250_000, 100
mask = pd.DataFrame(np.random.randint(0, 1, size=shape), dtype=pd.BooleanDtype())
%timeit mask.transpose()
# 4.28 s ± 12.3 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)  # <-- main
# 1.12 s ± 19.6 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)  # <-- PR

[rhshadrach]

@jorisvandenbossche - I don't understand this TODO, can you give more details?

[jorisvandenbossche]

Hmm, I suppose that with the currently implementation, the resulting arrays are never views but always copies? (since result_data and result_mask are newly created arrays (and then the values are copied in)
In that case I think you can just remove the comment

[jorisvandenbossche]

Ah, that was just copied from the as_array implementation above, then can certainly be removed. Sorry for the confusion!

[jbrockmendel]

I've been thinking about this and #52083. This is a workaround for lack of 2D support in MaskedArray (not unlike #52261). Maybe its time to revisit actually supporting 2D in MaskedArray?

[rhshadrach]

I've been thinking about this and #52083. This is a workaround for lack of 2D support in MaskedArray (not unlike #52261). Maybe its time to revisit actually supporting 2D in MaskedArray?

Agreed this is a workaround, no opposition to 2D for masked dtypes.

[github-actions]

This pull request is stale because it has been open for thirty days with no activity. Please update and respond to this comment if you're still interested in working on this.