Operationalize and Monitor Models SNOWPRO-ADVANCED-DATA-SCIENTIST Practice Questions (Page 1)

Questions 1–5

1. 1foundation · easy

   In a production ML system, which metric is most appropriate for detecting that the model's inference speed has degraded, potentially affecting user experience?

   • APrediction accuracy
   • BModel latency
   • CData drift
   • DModel version
   Reveal Solution Discussion Select an answer first
2. 2expert · hard

   A fintech company has a loan approval model in production. The model is updated monthly. After a recent update, the model's default rate increased, but the approval rate also increased. The team is considering rolling back to the previous version. However, the previous version had a lower approval rate, which could reduce revenue. What is the best decision?

   • ARoll back to the previous version and also reduce the approval rate further to minimize risk.
   • BRoll back to the previous version immediately to reduce default risk.
   • CAnalyze the trade-off between default rate and approval rate, and consider adjusting the decision threshold or retraining before deciding.
   • DKeep the new model and accept the higher default rate as a cost of higher approvals.
   Reveal Solution Discussion Select an answer first
4. 3foundation · easy

   When should a model rollback be executed?

   • AWhen a new model version is released
   • BWhen the model's training data is updated
   • CWhen the model's latency is slightly higher than expected
   • DWhen the new model version shows a significant performance drop in production
   Reveal Solution Discussion Select an answer first
5. 4application · medium

   A retail bank has a model that predicts customer lifetime value. The data science team wants to create a dashboard for the executive team to show the model's business impact. Which metric should be prominently displayed to demonstrate business value?

   • AThe model's accuracy on the test set.
   • BThe estimated increase in revenue or cost savings attributed to the model.
   • CThe number of predictions made per day.
   • DThe model's precision and recall.
   Reveal Solution Discussion Select an answer first
6. 5foundation · easy

   Which situation most strongly indicates that a deployed model should be retrained?

   • AA change in the model's inference latency
   • BA single prediction error observed in production logs
   • CA sustained drop in model accuracy below an acceptable threshold
   • DA new version of the programming language used for the model code
   Reveal Solution Discussion Select an answer first