[Python] pandas grouper + time shifting

Here my pandas:

d= {'dates': ['2020-07-16','2020-07-15','2020-07-14','2020-07-13','2020-07-12','2020-07-11','2020-07-10'],
'T':[3392,3020,663,2877,366,200,100], 'P': [53,91,0,55,3,22,55]}
df = pandas.DataFrame(data=d)
df['dates']=pandas.to_datetime(df['dates'])


If I pack my pandas each 3 days using Grouper:

for _, grp in df.groupby(pandas.Grouper(key='dates', freq='3D')):#
print (grp)
dates T P
6 2020-07-10 100 55
5 2020-07-11 200 22
4 2020-07-12 366 3
dates T P
3 2020-07-13 2877 55
2 2020-07-14 663 0
1 2020-07-15 3020 91
dates T P
0 2020-07-16 3392 53


Now I want to pack each 3 days but with one days shift:

8 2020-07-10 100 55
7 2020-07-11 200 22
6 2020-07-12 366 3
dates T P
11 2020-07-11 200 22
10 2020-07-12 366 3
9 2020-07-13 2877 55
dates T P
8 2020-07-12 366 3
7 2020-07-13 2877 55
6 2020-07-14 663 0
dates T P
5 2020-07-13 2877 55
4 2020-07-14 663 0
3 2020-07-15 3020 91
dates T P
2 2020-07-14 663 0
1 2020-07-15 3020 91
0 2020-07-16 3392 53


How can I stipulate this to Grouper ?

Continue reading...