xgboost

Definition

A widely used library implementing gradient-boosted decision trees, with regularization, built-in handling of missing values, and efficient parallel tree construction. It became the default strong baseline for tabular prediction and is the bar deep-learning papers must clear to claim progress on structured data. Note that it is one implementation, not the algorithm: gradient boosting predates it, and LightGBM and CatBoost are peers.

Explainers using this term

• Decision Trees and Gradient Boosting — Still the Champion on Tabular DataXGBoost: A Scalable Tree Boosting System