Sort the response actions
Select response categories that fit common model monitoring findings.
Match each monitoring finding to the response action that best fits first. Rollback Recalibrate Retrain Investigate Safeguard New model version increases latency and fails the canary guardrail A deployed parser maps valid partner categories to missing Ranking order is stable, but predicted probabilities are consistently too high Mature labels show the target relationship changed after a permanent policy shift A medium-importance feature drifts, but outputs and guardrails are stable A high-stakes low-volume slice shows possible degradation with wide uncertainty Borderline automated decisions may harm customers while labels mature Score distribution jumps without known deploy or feed change
Sign up free — one personalized lesson every day, matched to your role and goals.
Already have an account? Sign in