News for hackers. Stuff that matters.

A comment on CatBoost: Open-source gradient boosting library. Parent comment

Re:CatBoost: Open-source gradient boosting library by knightoffaith on Wednesday March 06, 2024 @08:07PM (#39623875)
Interesting - did this work better than a single model? In general, do meta-ensembles work better? My sense was that just xgboost was the main winner of kaggle competitions.
[ Parent | on HN ]
Re:CatBoost: Open-source gradient boosting library by intelkishan on Thursday March 07, 2024 @12:24AM (#39625331)
Meta-ensembles are the new trend for tabular data on Kaggle
[ Parent | on HN ]
Re:CatBoost: Open-source gradient boosting library by anmorgan24 on Thursday March 07, 2024 @04:13PM (#39634655)
https://mlcontests.com/state-of-competitive-machine-learning...
[ Parent | on HN ]
Re:CatBoost: Open-source gradient boosting library by hcarlens on Thursday March 07, 2024 @05:02AM (#39627167)
XGBoost, LightGBM, and Catboost are all used quite frequently in competitions. LightGBM is actually marginally more popular than the other two now, but it's pretty close. In the M5 forecasting competition a few years back, many of the top solutions used primarily LightGBM.
[ Parent | on HN ]
Re:CatBoost: Open-source gradient boosting library by shiandow on Thursday March 07, 2024 @07:04AM (#39628045)
If you're clever about how you blend models you can pretty much ensure that the performance of the (weighted) averaged model is strictly better than the (weighted) average performance of the models.

And increasing the space of possible models pretty much guarantees the performance will improve, provided you can still find a good enough optimum.

[ Parent | on HN ]