Metadata-Version: 2.4
Name: explainbench
Version: 0.1.1
Summary: A toolkit for interpretable machine learning and fairness auditing
Home-page: https://github.com/jamesafful/explainbench
Author: James Afful
Author-email: affulj@iastate.edu
Classifier: Programming Language :: Python :: 3
Classifier: License :: OSI Approved :: MIT License
Classifier: Operating System :: OS Independent
Requires-Python: >=3.8
Description-Content-Type: text/markdown
Requires-Dist: numpy<1.27,>=1.24
Requires-Dist: pandas==1.5.3
Requires-Dist: scikit-learn==1.3.2
Requires-Dist: matplotlib==3.7.1
Requires-Dist: shap==0.44.1
Requires-Dist: lime==0.2.0.1
Requires-Dist: dice-ml==0.11
Requires-Dist: streamlit==1.27.2
Requires-Dist: typing-extensions>=4.5.0
Dynamic: author
Dynamic: author-email
Dynamic: classifier
Dynamic: description
Dynamic: description-content-type
Dynamic: home-page
Dynamic: requires-dist
Dynamic: requires-python
Dynamic: summary

# Explainbench: An Open-Source Toolkit for Interpretable Machine Learning

**Explainbench** is a Python toolkit that makes powerful ML interpretability techniques like SHAP, LIME, counterfactuals, and global surrogate models accessible and usable — especially for high-stakes, public-sector applications.

## Features

- **Unified Interface** for SHAP, LIME, and DiCE
- **Fairness & Explainability Metrics** (Disparate Impact, Fidelity, Consistency)
- **Preloaded Datasets** (COMPAS, Adult Income, etc.)
- **Interactive Visualizations** with Streamlit and Plotly
- **Notebook Examples** for quick understanding and classroom use

## Why It Matters

As ML systems are increasingly used in criminal justice, healthcare, and finance, it's crucial that we can explain, audit, and challenge their decisions. `Explainbench` provides transparent tools for evaluating black-box models in real-world, socially relevant contexts.

## Installation

```bash
pip install explainbench
