A monitoring dashboard shows that AUC-ROC for a binary classifier has remained above 0.85 but statistical parity difference has deteriorated to -0.22 over the past 30 days. What does this combination of signals indicate?
Select an answer to reveal the explanation.
Short Explanation and Infographic
Think of it like setting up smoke detectors in a building — you configure them before anything can go wrong. AUC-ROC above 0.85 indicates strong model discriminative ability with no quality degradation, while a statistical parity difference of -0.22 reveals the monitored group receives favorable outcomes at a meaningfully lower rate than the reference group. watsonx.governance tracks quality and fairness as separate monitors because a model can achieve high overall accuracy while exhibiting demographic bias in its outcome distribution.
Full explanation below image
Full Explanation
AUC-ROC above 0.85 indicates strong model discriminative ability with no quality degradation, while a statistical parity difference of -0.22 reveals the monitored group receives favorable outcomes at a meaningfully lower rate than the reference group. watsonx.governance tracks quality and fairness as separate monitors because a model can achieve high overall accuracy while exhibiting demographic bias in its outcome distribution. Both signals together provide a complete view of the model's governance posture. The incorrect options — such as 'The model is experiencing data drift that has simultaneously degraded both predictive performance and demographic fairness' and 'The model's overall discriminative ability is declining and statistical parity difference is a secondary indicator of model quality drift' — describe either out-of-sequence steps or unrelated configuration tasks. This concept falls under the 4.0 Configure Evaluation and Monitoring domain of the IBM watsonx.governance certification.