A production model's KS statistic drops from 0.65 to 0.41 over 6 months. PSI for input features remains below 0.10. What is the most likely explanation?
Select an answer to reveal the explanation.
Short Explanation and Infographic
Here's the deal — b is correct because stable PSI (< 0.10 for features) indicates input distributions have not shifted, yet discriminative power (KS) has declined. This pattern points to concept drift: P(Y|X) has changed — the relationship between features and target has evolved.
Full explanation below image
Full Explanation
B is correct because stable PSI (< 0.10 for features) indicates input distributions have not shifted, yet discriminative power (KS) has declined. This pattern points to concept drift: P(Y|X) has changed — the relationship between features and target has evolved. A is wrong because feature PSI is stable. C (implementation drift) would typically cause a sudden change, not gradual decline. D (label drift) would affect ground truth labels in the monitoring evaluation.