02_end_to_end_machine_learning_project - LightGBM model not working

[2Nerds1BlackHole]

I had trouble installing lightgbm v3.2.1 (as suggested in the environment .yml) so I installed the latest I could, version 4.3.0. When I got to the lgb.train function in the book, the "early_stopping_rounds=200" parameter is used, but this parameter no longer (directly) exists in that function.

Omitting that parameter yields terrible results, but I found a compatibility fix on the LightGBM github repo. They removed this parameter and put it in something called callbacks.

I replaced "early_stopping_rounds=200" parameter with "callbacks=[lgb.early_stopping(stopping_rounds=200)]" and it worked just fine with results very similar to the book.

First time using github, so apologies if this doesn't go here. Hopefully this helps someone encountering the same issue!