ML//gradient boosting

Ensemble that builds many shallow decision trees in sequence, each correcting the residual errors of the ones before it.

Ensemble that builds many shallow decision trees in sequence, each correcting the residual errors of the ones before it.

Dominant on tabular data: exploits nonlinear interactions between features without learning a representation from scratch.

Often the real rival to an LSTM on small or medium tabular time-series problems, and frequently wins there.

Implementations: XGBoost, LightGBM, CatBoost. Prone to overfitting without regularization and early stopping.