LIME

LIME (Local Interpretable Model-agnostic Explanations) is an approach for explaining the predictions of machine learning models by approximating them locally.