ML Interpretability Cheat Sheet

Last Updated: November 21, 2025

Technique Comparison

Method Strength
SHAP Consistent local contributions
LIME Sample-based perturbations
Partial Dependence Global marginal effects

Commands

shap.TreeExplainer(model)
Compute SHAP values
LimeTabularExplainer(...)
Wrap dataset for LIME
plot_partial_dependence
Visualize feature curves

Fairness & Guardrails

Monitor disparate impact, audit training data, and document decisions for compliance.

💡 Pro Tip: Blend global explanations (feature importance) with local (per-sample) insights.
← Back to Data Science & ML | Browse all categories | View all cheat sheets