Allow diff & shift to take lists of periods?

[TomAugspurger]

I've seen code like

In [36]: s = pd.Series(range(10))

In [37]: pd.concat([s.diff(i).rename(f"L{i}") for i in range(10)], axis=1)
Out[37]:
    L0   L1   L2   L3   L4   L5   L6   L7   L8   L9
0  0.0  NaN  NaN  NaN  NaN  NaN  NaN  NaN  NaN  NaN
1  0.0  1.0  NaN  NaN  NaN  NaN  NaN  NaN  NaN  NaN
2  0.0  1.0  2.0  NaN  NaN  NaN  NaN  NaN  NaN  NaN
3  0.0  1.0  2.0  3.0  NaN  NaN  NaN  NaN  NaN  NaN
4  0.0  1.0  2.0  3.0  4.0  NaN  NaN  NaN  NaN  NaN
5  0.0  1.0  2.0  3.0  4.0  5.0  NaN  NaN  NaN  NaN
6  0.0  1.0  2.0  3.0  4.0  5.0  6.0  NaN  NaN  NaN
7  0.0  1.0  2.0  3.0  4.0  5.0  6.0  7.0  NaN  NaN
8  0.0  1.0  2.0  3.0  4.0  5.0  6.0  7.0  8.0  NaN
9  0.0  1.0  2.0  3.0  4.0  5.0  6.0  7.0  8.0  9.0

Could we simplify that by allowing s.diff(range(10), prefix="L")?

How would we handle DataFrames? Prefix with the column name and a separator, like get_dummies?

[tliu30]

Hi! Would love to make this a reality...

I took a first look at how this could work for diff for Series, and it seems pretty simple (will attach a possible solution in a pull request shortly).

For DataFrames...

• when axis=0 i think we can get away with the same hack (also in pull request)
• when axis=1...maybe we just don't allow them to use multiple periods if axis=1?

(for shift, could apply a similar if-else block for pandas.core.generic.NDFrame.shift)

[rhshadrach]

Closing in favor of #44424