Pandas pct_change() using time series and preserving ids

Question

I am trying to calculate the perc change occurring between two dates of the values for each id in my df:

date        id  a  b
2021-01-01  1   6  4
2021-01-01  2