ENH: built-in shift function that shifts each row by a different period, given by a Series.

[glorencini]

Is your feature request related to a problem?

No

Describe the solution you'd like

I would like a built-in shift function that shifts each row by a different period, given by a Series.

API breaking implications

It would improve the shift possibilities.

Describe alternatives you've considered

Using shift with apply and lambda.

Additional context

[add any other context, code examples, or references to existing implementations about the feature request here]

def shift_df_by_series(self, df=pd.DataFrame, s=pd.Series) -> pd.DataFrame:
    return df.apply(lambda shifted_df: shifted_df.shift(periods=s[shifted_df.name], fill_value=0), axis='columns')

[phofl]

Hi,

if I understand you correctly, this should basically be a wrapper around apply specifically for shift?

Please give your issue a title

[glorencini]

Yes and no.
We don`t know if it is the best possible implementation, but it works. It works for our case, but it is not generic.
Also, their is no validation around the indexes and the method arguments. For example, it would be possible to shift only the indexes that are informed rather than result in error when an index is not passed.
I think it would be very useful to have a generic and error safe implementation of this shift functionality.

[jbrockmendel]

i think DataFrame.shift now supports a list of periods. Does that handle this use case?