DOC: DataFrame on df.nunique causing confusion

[regmibijay]

https://pandas.pydata.org/docs/reference/api/pandas.DataFrame.nunique.html

Documentation problem

The above documentation provides an example using pd.DataFrame({'A': [1, 2, 3], 'B': [1, 1, 1]}) upon which df.nunique() returns

A    3
B    1

This example might be a bit misleading as there are 3 and 1 present in dataframe as column elements and cause confusion.

Suggested fix for documentation

I would like to suggest using something like pd.DataFrame({'A': [5, 6, 7], 'B': [5, 1, 1]}) upon which df.nunique() returns

A 3
B 2

as this examples serves the purpose of nunique better as its clear that it counts the number of unique elements in column and does not return anything resembling any column element.

Furthermore, I would like to extend the example dataframe to df.nunique(axis=1) and match corresponding values with new dataframe as :

df.nunique(axis=1)
0    1
1    2
2    2
dtype: int64

Does this change need to be documented in whatsnew?

[mzeitlin11]

Thanks for the report @regmibijay! Those changes make sense to me and would definitely be an improvement. A documentation improvement wouldn't need a whatsnew.

[shalarewicz]

take