Exact Integrated Gradients for tree ensembles, including XGBoost, LightGBM, Random Forests, and Gradient Boosting. Computes fast, exact feature attributions for tree models without numerical gradients, numerical integration, Monte Carlo sampling, or path approximations.
Feature attribution for realized model fit (rather than predictions). EDEF provides exact additive decomposition of predictive performance for linear, PyTorch, JAX, and tree-based models, including standard errors, grouped inference, and exact TreeIG decompositions for tree-based models.
