DOC: Inconsistent use of "column-wise" and "row-wise"

[ianozsvald]

5 Pandas docstrings discuss the axis argument with one use of row/column-wise, 2 docstrings use an opposite meaning (and possibly more examples exist, I picked these up using Google & DuckDuckGo).

I believe the first 5 usages are in agreement with the Numpy usage.

Examples of axis: Equals 0 or ‘index’ for row-wise, 1 or ‘columns’ for column-wise (i.e. 0 is row-wise):

• https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.quantile.html
• https://pandas.pydata.org/docs/reference/api/pandas.DataFrame.swaplevel.html
• https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.idxmin.html
• https://pandas.pydata.org/docs/reference/api/pandas.DataFrame.idxmax.html
• https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.nunique.html

Examples of The axis to use. 0 or ‘index’ to compute column-wise, 1 or ‘columns’ for row-wise (i.e. 0 is column-wise):

• https://pandas.pydata.org/docs/reference/api/pandas.DataFrame.corrwith.html
• https://pandas.pydata.org/pandas-docs/version/0.24.2/reference/api/pandas.Series.all.html

For Series.all the examples down the page note Default behaviour (i.e. axis=0) checks if column-wise values all return True. and Specify axis='columns' (or axis=1) to check if row-wise values all return True..

On Numpy I can't find many examples, the few that I find are consistent with the first (5 case) usage:

• https://numpy.org/doc/stable/reference/generated/numpy.hsplit.html#numpy.hsplit horizontally (column-wise)...equivalent to split with axis=1
• https://numpy.org/doc/stable/reference/generated/numpy.vsplit.html#numpy.vsplit vertically (row-wise)... equivalent to split with axis=0 (default)

Suggested fix for documentation

I think that the 2 opposing cases (corrwith and Series.all) don't need to have the inverted explanation so I think that reversing them to fit the general case makes sense, but maybe I'm missing something?

[mattharrison]

On a semi-related note (since this isn't the axis parameter), the behavior of the normalize parameter for pd.crosstab also accepts 'columns' or 'index'. When you pass 'columns', it doesn't normalize across the columns, but down each one. I understand the behavior, but it seems inconsistent. 🤷‍♀️😢

[jackgoldsmith4]

take