A deployed scoring model accumulates 30 days of production predictions. PSI = 0.22 is detected for the predicted score distribution. According to standard PSI thresholds, what action should be taken?
Select an answer to reveal the explanation.
Short Explanation
Here's the deal — when PSI hits 0.22, you're past the 0.20 significant-shift threshold, so you escalate for model review, dig into root cause, and consider retraining or replacement.
Full Explanation
Standard PSI thresholds are: PSI < 0.10 = negligible change, 0.10-0.20 = moderate change (monitor closely, investigate), and PSI >= 0.20 = significant change. PSI = 0.22 exceeds the 0.20 threshold, so the right action is to escalate for model review, investigate root cause, and consider retraining or replacement. The claim that 0.22 is negligible because it is below 0.25 is wrong because 0.22 is above 0.20. The claim that 0.22 falls in the moderate range is wrong because it exceeds that range. The claim that SAS Model Manager automatically rolls back to the previous champion is wrong because it can alert but does not auto-rollback.