Explainer: SHAP Tree

class xaicompare.adapters.explainers.explainer_shap_tree.ShapTreeExplainerAdapter(model_adapter, config: Dict[str, Any])

Bases: ExplainerAdapter

SHAP Tree Explainer with:

• Safe sparse handling

• Forced vectorization

• Guaranteed 2-D input to SHAP

• Batch processing to avoid memory errors

• Multi-class normalization

_abc_impl = <_abc._abc_data object>

global_importance(X, rows_limit: int = 200) → Tuple[ndarray, List[str]]

Compute mean|SHAP| across first rows_limit samples safely.

limitation_text()

local_explanations(x_row) → ndarray

Compute SHAP for a single example → return signed vector.