pooled rolling window regression in python

Question

I have a dataframe of the following structure:

id t  y   x
1  1  20  2
1  2  25  1
1  3  21  2
1  4  32  3
2  3  35  7