Module explainable 

Explainability wrappers for inference monitoring Explainable AI Integration for Inference Monitoring

This module provides native Explainable trait implementations for APR format models, with decision path types for full prediction traceability.

Toyota Way: 現地現物 (Genchi Genbutsu)

Every prediction can be traced to its decision path. All decisions are explainable.

Features

• LinearExplainable: Wraps linear models with feature contribution tracking
• TreeExplainable: Wraps decision trees with split path tracking
• EnsembleExplainable: Wraps ensembles with per-model aggregation

Example

use aprender::linear_model::LinearRegression;
use aprender::explainable::LinearExplainable;
use aprender::explainable::path::{Explainable, DecisionPath};

let model = LinearRegression::new();
model.fit(&x, &y)?;

// Wrap with explainability
let explainable = LinearExplainable::new(model);
let (outputs, paths) = explainable.predict_explained(&features, 1);
println!("{}", paths[0].explain());

Modules

path

Decision Path Types and Explainability Trait

Structs

EnsembleExplainable

Wrapper that makes RandomForestRegressor explainable for inference monitoring.

LinearExplainable

Wrapper that makes LinearRegression explainable for inference monitoring.

LogisticExplainable

Wrapper that makes LogisticRegression explainable for inference monitoring.

TreeExplainable

Wrapper that makes DecisionTreeRegressor explainable for inference monitoring.

Traits

IntoEnsembleExplainable

Extension trait to easily convert RandomForestRegressor to explainable.

IntoExplainable

Extension trait to easily convert LinearRegression to explainable.

IntoLogisticExplainable

Extension trait to easily convert LogisticRegression to explainable.

IntoTreeExplainable

Extension trait to easily convert DecisionTreeRegressor to explainable.