Agent skill · research science · aj-geddes
ML Model Explanation
Interpret machine learning models using SHAP, LIME, feature importance, partial dependence, and attention visualization for explainability
Why this skill is useful
Provides specific techniques and Python implementations for interpreting machine learning models that the AI wouldn't reliably generate on its own.
What it needs
Requires matplotlib, numpy, pandas, scikit-learn, seaborn installed locally. About 5k tokens when loaded. Last updated 2026-03-04. 307 stars on the source repository.
What this skill does
ML Model Explanation Model explainability makes machine learning decisions transparent and interpretable, enabling trust, compliance, debugging, and actionable insights from predictions. Explanation Techniques Feature Importance: Global feature contribution to predictions SHAP Values: Game theory-based feature attribution LIME: Local linear approximations for individual predictions Partial Dependence Plots: Feature relationship with predictions Attention Maps: Visualization of model focus areas Surrogate Models: Simpler interpretable approximations Explainability Types Global: Overall model behavior and patterns Local: Explanation for individual predictions Feature-Level: Which features matter most Model-Level: How different components interact Python Implementation Explanation Techniques Comparison Feature Importance: Fast, global, model-specific SHAP: Theoretically sound, game-theory based, computationally expensive LIME: Model-agnostic, local explanations, interpretable PDP: Shows feature relationships, can be misleading with correlations Attention: Works for neural networks, interpretable attention weights Interpretability vs Accuracy Trade-off Linear models: Highly interpretable, lower accuracy Tree models: Interpretable, moderate accuracy Neural networks: High accuracy, less interpretable Ensemble models: High accuracy, need explanation techniques Regulatory Compliance GDPR: Right to explanation for automated decisions Fair Lending: Explainability for credit decisions Insurance: Transparency in underwriting Healthcare: Medical decision explanation Deliverables Feature importance rankings Local explanations for predictions Partial dependence plots Global behavior analysis Model interpretation report Explanation dashboard
How to use it
Reference it in AdaL, Claude Code, Cursor or any coding agent — nothing to install:
@skills aj-geddes/ml-model-explanation