This code is mainly based on the baseline from Leocd - https://gitlab.aicrowd.com/leocd/zew-data-purchasing-challenge-2022-starter-kit/-/blob/master/run.py. Please give him a vote ! It scored 0.877
I have done a tiny modification - train on all data. No data left for a real validation. It scored 0.885.
The submission is here https://gitlab.aicrowd.com/moto/data-purchasing-hello/-/issues/10.
Well, if we follow the best startegy in this notebook https://www.aicrowd.com/showcase/ways-to-select-which-data-to-purchase-episode-1 we could reach the top position in round 1 ;-)
Good luck, everyone !