Today I applied to a kaggle competition, it's actually a college project. Training a ML model.
Some fields are categorical but the data set is large, above 100K. I choose to see who'll do better - XGBoost, LGBM or CatBoost.
XGBoot lost immediately, I used F1 Macro as my evaluation an I got 0.70.
In LGBM it reached till 0.73 but I'll need to fine tune it since in leaders it only got till 0.67.
Now I'm doing CatBoost and it's taking too much time. 🥲
Top comments (0)