im trying to apply np.roll with groupy of pandas.

Actually, i have this dataframe: (In the columns complete_contracts_shift i applied shift with groupby:

df.groupby(['contracts ','param_contrct'])['complete_contracts '].shift(1)

I need that dataframe has the column 'complete_contracts_shift' as follows:

i know np.roll of numpy, but i cant combinate this with groupby.

CodePudding user response：

If you are getting this result and consistently missing the third month. I suggest to go for a more simplistic approach with:

df['complete_contracts_shift'] = np.where(df['complete_contracts_shift'].isna(),df['contracts'] '-' df['param_contrct'] '-03/01/2022',df['complete_contracts_shift']

This will help you fill the NaN values with the correct combination of columns and date values.

The data you provide is pitifully not enough for me to give more a detailed answer.

CodePudding user response：

I could solve it with this code!

    _df['complete_contracts_shift'] = _df.groupby(['contracts','param_contrct'])['complete_contracts'].transform(lambda x: x.values[::-1])