CatBoost - the new generation of gradient boosting

on YouTube ↗38m

" summary

Claim: CatBoost outperforms other GBM libraries on public benchmarks and ships with concrete implementations: symmetric trees for very fast prediction, permutation-based categorical statistics to avoid target leakage, and a relaxed order-boosting scheme that reduces overfitting on small datasets. Benchmarks show ~20× GPU training speedups versus XGBoost and 30–60× faster prediction in latency‑sensitive settings; code, reproducible plots and tuned comparisons are available on github.

AI-written summary · about

" chapters (13)

1. 0:00CatBoost introduction
2. 3:19Gradient boosting basics
3. 5:29Why CatBoost excels
4. 8:08Categorical features handling
5. 15:37Unbiased gradient estimation
6. 18:27Training speed benchmarks
7. 19:31GPU training
8. 21:40Fast prediction
9. 22:01Model insights
10. 27:05Overfitting detection
11. 31:02Key tuning parameters
12. 34:33Research papers
13. 35:38Q&A

" related talks

• 01h 27mCatBoost - The New Generation of Gradient BoostingVasily Ershov · PyData Los Angeles 2018
• 01h 33mCrunching Your Data with CatBoost - the New Gradient Boosting LibraryVasily Ershov · PyData NYC 2018
• 40mCatBoost: Fast Open-Source Gradient Boosting Library For GPUVasily Ershov · PyData Berlin 2018
• 01h 28mMastering gradient boosting with CatBoostAnna Veronika Dorogush · PyData London 2019
• 37mGradient Boosting for Data with Both Numerical and Text FeaturesVasily Ershov · PyData Los Angeles 2019
• 18mCompetitive Analysis of the Top Gradient...Sai Ayachit, Shyam R, Anubhav Singh, Vinayak Patil · OpenPOWER Summit NA 2020
• 42mHow am I going to deal with all of these catsLiam Kirwin · PyData London 2019
• 41mBetter Than Deep Learning: Gradient Boosting Machines (GBM)Szilard Pafka · Berlin Buzzwords 2019