What is TreeSHAP?

TreeSHAP

TreeSHAP is a fast explainer used for analyzing decision tree models in the Shap python library. Generally, it is considered useful for XGBoost or LightGBM.

TreeSHAP is designed for tree-based machine learning models such as decision trees, random forests and gradient boosted trees. TreeSHAP is offered as a rapid, model-specific alternative to KernelSHAP; however, it can sometimes produce unintuitive feature attributions. Given the prevalence of XGBoost, TreeSHAP is an explainer approach heavily used in various teams. The tree explainer makes use of the structure of the tree algorithm to extract and generate SHAP values much faster than the Kernel explainer.