UserWarning: `DataFrame.groupby_on_multiple_columns` defaulting to pandas implementation


d[‘column1’] = d[(d.merchant_count==0) & (d.amount<-100)].groupby([‘card_id’, ‘year’, ‘month’, ‘week’]).cumcount().astype(np.float16)+1

UserWarning: DataFrame.groupby_on_multiple_columns defaulting to pandas implementation.

Will groupby on multiple columns be added soon?

Hi @davidjhp, thanks for the question! Right now we have a PR #2461 which is adding a possibility of performing groupby on multiple columns operation for aggregation functions ( cumcount is included in this scope). We are trying to complete this activity by the EOY.

Hi @davidjhp, the PR has been merged into master.