MLens v0.2.0 — Production Ready 🚀

✨ What's New in v0.2.0

🖥️ CLI Tool — Run audits from terminal

mlens audit model.pkl X_test.csv y_test.csv
mlens audit model.pkl X_test.csv y_test.csv --sensitive gender.csv
mlens version

📊 Streamlit Live Dashboard

streamlit run dashboard/app.py

📄 PDF Report Export

report.save("audit_report.pdf")

🧪 Tests Added

• test_auditor.py — 11 unit tests
• test_fairness.py — 9 unit tests
• test_drift.py — 10 unit tests

📄 HTML Audit Report

• Full interactive report with SHAP, Fairness & Drift charts
• Works offline, no server needed

🐛 Bug Fixes

• Fixed KS-test crash on constant features
• Fixed SHAP dual-output for binary classifiers

⬆️ Upgrade

pip install mlens --upgrade

📊 Release Stats

• 3 major new features
• 10 new files added
• 30 unit tests

MLens v0.1.0 — Initial Release

🚀 First release of MLens — Explainable ML Audit Tool

Features:

• 🧠 SHAP explainability (Tree/Linear/Kernel auto-selection)
• ⚖️ Fairness evaluation (Demographic Parity, Equalized Odds)
• 📊 Drift detection (PSI + KS-test per feature)
• 📄 HTML audit report generation