5 / 6 Interpretability

Zoom In, Zoom Out

A team explains a fraud-detection model in two ways: one report shows how the model behaves across the entire customer base on average, and a second report shows why one specific customer's transaction was flagged. Which pairing correctly labels these two explanation types?

A

Both are global explanations, just at different aggregation levels

B

The first is a counterfactual explanation, the second is a global explanation

C

The first is a global explanation, the second is a local explanation

D

The first is a local explanation, the second is a global explanation

E

Both are local explanations, since both come from the same trained model