771: Gradient Boosting: XGBoost, LightGBM and CatBoost, with Kirill Eremenko
Super Data Science: ML & AI Podcast with Jon Krohn
00:00
Exploring Sparse Columns and Efficient Decision Tree Building with Gradient Boosting
The chapter delves into the use of sparse columns in analytics projects, highlighting their significance in various industries like ore mining, medical datasets, and time sheets. It discusses the strategies employed by LightGBM, including gradient-based one-side sampling and leaf-wise tree growth, to enhance the speed of decision tree building while sacrificing minimal accuracy. The chapter also explores CatBoost's advantages in handling categorical variables through target encoding, avoiding data leakage, and utilizing symmetric trees for efficient decision-making processes.
Transcript
Play full episode
