Struct InterventionalTreeExplainer 

pub struct InterventionalTreeExplainer<'a> { /* private fields */ }

Exact interventional TreeSHAP using an explicit background distribution. Unlike TreeExplainer, absent features are replaced from background rows rather than integrated using training-path covers.

Implementations

impl<'a> InterventionalTreeExplainer<'a>

pub fn new(model: &'a TreeEnsemble, background: Background) -> Self

pub fn with_max_features(self, max_features: usize) -> Self

pub fn explain_probability(&self, x: ArrayView2<'_, f64>) -> Result<Explanation>

Explains probabilities exactly under the supplied background. A one-output ensemble uses sigmoid; multiple outputs use softmax.

pub fn explain_binary_log_loss( &self, x: ArrayView2<'_, f64>, targets: &[bool], ) -> Result<Explanation>

Explains binary logistic loss for each sample’s target label.

Trait Implementations

impl Explainer for InterventionalTreeExplainer<'_>

fn explain(&self, x: ArrayView2<'_, f64>) -> Result<Explanation>