I have a df:
df = Date id1 amount is_winner
2022-07-14 02:34:20.348. A. 87.11. False
2022-07-14 02:34:20.348. B. 77.12. True
2022-07-14 02:37:20.348. A 89.11. False
2022-07-14 02:37:20.348. B. 87.12. True
2022-07-14 02:44:20.348. A. 81.11. True
2022-07-14 02:44:20.348. B. 87.12. False
2022-07-14 02:54:20.348. A. 99.11. False
2022-07-14 02:54:20.348. B. 99.11. True
Now, I want to apply groupby(["date", "id1"]) and have various measures that are based on previous groups (notice data is sorted by date):
df =
Date id1 amount is_winner delta_amount_from_last_win num_wins_before. num_groups_before. rank_in_prev_group. time_from_prev(minutes)
2022-07-14 02:34:20.348. A. 87.11. False. 0. 0. 0. nan. nan
2022-07-14 02:34:20.348. B. 77.12. True 0. 0. 0. nan. nan
2022-07-14 02:37:20.348. A 89.11. False. 2. 0. 1. 1. 3
2022-07-14 02:37:20.348. B. 87.12. True. 10 1. 1. 0. 3
2022-07-14 02:44:20.348. A. 81.11. True. -8. 0. 2. 1. 7
2022-07-14 02:44:20.348. B. 87.12. False. 0. 1. 2. 0. 7
2022-07-14 02:54:20.348. A. 99.11. False. 18. 1 3. 0. 10
2022-07-14 02:54:20.348. B. 99.12. True. 12 2. 3. 1. 10
How can it be done efficiently?