DOC: Add example for pandas.DataFrame.rolling() with on

[luke396]

• closes DOC: No example for on kwarg of DataFrame.rolling() #50080

I'm a newbie, and welcome any comments!

[phofl]

It looks like the docstring validation is failing. Could you investigate?

I think one example is sufficient

[luke396]

It looks like the docstring validation is failing. Could you investigate?

I think one example is sufficient

Ok, I will try to pass the docstring validation, and reduce two example to one.

Thanks for your comments!

[luke396]

When I use python make.py --single pandas.DataFrame.rolling, there is a warning, WARNING: extlinks: Sphinx-6.0 will require a caption string to contain exactly one '%s' and all other '%' need to be escaped as '%%'., I'm not sure that matters.

[luke396]

I'm so sorry that my improper operation messed up the PR!

[MarcoGorelli]

wouldn't this give the same result even without on='A'? might be more illustrative to have an example in which on= makes a difference

[luke396]

Without on='A', the result will be like this:

>>> df.rolling(2).sum()
     A     B
0  NaN   NaN
1  4.0   6.0
2  8.0   NaN
3  NaN   NaN
4  NaN  14.0

Do I need to add this result as comparison?

[MarcoGorelli]

I was thinking more something like

In [62]: df = pd.DataFrame({
    ...:     'A': to_datetime(['2020-01-01', '2020-01-01', '2020-01-02']),
    ...:     'B': [1,2,3],
    ...:     },
    ...:     index=date_range('2020', periods=3))

in which if you do rolling('D'), then the values of 'B' differ if you do on='A' (instead of the default, which uses the index)

[luke396]

Oh, I see what that means. How about the following:

>>> df = pd.DataFrame({
...     'A': [pd.to_datetime('2020-01-01'),
...           pd.to_datetime('2020-01-01'),
...           pd.to_datetime('2020-01-02'),],
...     'B': [1, 2, 3], },
...     index=pd.date_range('2020', periods=3))

>>> df
                    A  B
2020-01-01 2020-01-01  1
2020-01-02 2020-01-01  2
2020-01-03 2020-01-02  3

>>> df['B'].rolling('2D').sum()  # to avoid warning when sum on 'A'
2020-01-01    1.0
2020-01-02    3.0
2020-01-03    5.0
Freq: D, Name: B, dtype: float64

>>> df.rolling('2D', on='A').sum()   # value of 'B' is differ from above
                    A    B
2020-01-01 2020-01-01  1.0
2020-01-02 2020-01-01  3.0
2020-01-03 2020-01-02  6.0

[MarcoGorelli]

# value of 'B' differs from above

other than that, looks fine

[luke396]

Thanks for the correction, my English skills are a little rusty.

[MarcoGorelli]

Almost there 💪

[MarcoGorelli]

2 days?

[MarcoGorelli]

Seeing as you start with

Rolling sum with a window length of 2 on specific columon.

we can probably exclude this example. If someone tries running it locally, then can remove on= and see the difference, at which point it should be clear what it's done

[MarcoGorelli]

let's remove the comment

[luke396]

Almost there 💪

How about this one 😄

Thank you for taking the time to share your thoughts with me. I really appreciate your review.

[MarcoGorelli]

looks good, thanks @luke396 !