Question 940
After a WAN redesign, an anomaly model that worked before produces many false alarms. What lifecycle issue should be evaluated?
Explanation
Correct Answer
Input distribution changes can cause model drift and degrade accuracy.
Incorrect Answers
Generative models can synthesize human-readable explanations but require validation; for anomaly detection over-alerting after a WAN topology redesign, the relevant operation is detecting when a previously trained model stops representing current conditions, not drafting a human-readable network troubleshooting explanation.
Supervised learning can use historical labeled examples to learn outcome patterns; for anomaly detection over-alerting after a WAN topology redesign, the relevant operation is detecting when a previously trained model stops representing current conditions, not creating a supervised model from labeled incident data.
Predictive models infer patterns from examples to estimate a future or unknown outcome; for anomaly detection over-alerting after a WAN topology redesign, the relevant operation is detecting when a previously trained model stops representing current conditions, not estimating future events before an observed failure.