Implementation of gradient boosted decision trees, covering binary classification, regression, Poisson count, and multiclass objectives. It handles training with row and column subsampling, missing value routing, gamma pruning, and leaf regularization, then exposes prediction, probability, raw margin, feature importance, and full model serialization through a single XGB class.
Read docs/README.md for full API reference, objectives, importance types, inference, and persistence.
Note
Prerequisites: Deno for all development tasks.
Check - format, lint, and typecheck source:
deno task check- XGBoost - Original C++ project this port derives its algorithms from
- XGBoost Docs - Parameter reference and mathematical background
- Introduction to Boosted Trees - Formal derivation of gain, weight, and pruning