lore_sa - Local Rule-based Explanations

LORE (LOcal Rule-based Explanations) is a model-agnostic explanation method for black box classifiers. It provides interpretable explanations through decision rules, counterfactual scenarios, and feature importance scores.

Key Features:

• Model-agnostic: Works with any black box classifier

• Rule-based explanations: Natural IF-THEN rules

• Counterfactual reasoning: Shows “what-if” scenarios

• Feature importance: Identifies key decision factors

• Production-ready: Stable and actionable implementation

Quick Links:

• Paper: https://arxiv.org/abs/1805.10820

• GitHub: https://github.com/kdd-lab/LORE_sa

• Issues: https://github.com/kdd-lab/LORE_sa/issues

Getting Started

• Get started
  • Welcome to LORE_sa
  • What is LORE?
  • Installation
  • Quick Start
  • Key Components
  • Choosing an Explainer
  • Understanding the Explanation
  • Next Steps
  • Common Pitfalls
  • Getting Help
• Architecture and Methodology
  • Overview
  • Theoretical Foundation
  • The Four-Stage Process
  • Architecture Components
  • Workflow Diagram
  • Best Practices
  • Computational Complexity
  • References
  • Implementation Notes
• Tabular explanation example
  • Learning and explaining German Credit Dataset
  • Loading and preparation of data
  • Explaining the prediction
  • Learning a different model
  • Explaining the prediction

API Reference

Modules

• Modules
  • lore_sa.bbox: BlackBox abstract class
  • lore_sa.dataset: Dataset class
  • lore_sa.neighgen: Neighborhood Generator classes
  • lore_sa.discretizer: Discretizer classes and functions
  • lore_sa.encoder_decoder: Encoder/Decoder classes and functions
  • lore_sa.explanation: Explanation classes and functions
  • lore_sa.rule: Rule classes and functions
  • lore_sa.surrogate: Surrogate classes and functions
  • lore_sa.util: Util functions

Indices and tables

• Index

• Module Index

• Search Page