Oversee Model Governance PMI-CPMAI Practice Questions (Page 7)

Questions 31–34

1. 31expert · hard

   A manufacturing company has a governance policy that requires all model changes to be approved by a change advisory board (CAB). The team uses a CI/CD pipeline that automatically deploys models to production. A data scientist notices that the pipeline can deploy a model without CAB approval if the change is tagged as 'minor'. The governance officer wants to close this loophole. What is the most effective control?

   • ARequire the CI/CD pipeline to check for a CAB approval token before deploying any model, regardless of the change type.
   • BAsk the data scientist to manually review all changes before they are tagged as 'minor'.
   • CLog all 'minor' changes and review them at the next CAB meeting.
   • DRemove the 'minor' tag option from the CI/CD pipeline so all changes require CAB approval.
   Reveal Solution Discussion Select an answer first
2. 32foundation · easy

   In a documented model lifecycle, which stage involves systematically decommissioning a model that is no longer needed or has been replaced?

   • ARetirement
   • BMonitoring
   • CDeployment
   • DValidation
   Reveal Solution Discussion Select an answer first
4. 33application · medium

   A retail company is developing a new inventory-forecasting model. The team has completed the development and validation stages and is about to deploy to production. The operations manager wants to ensure the model can be monitored and updated without disrupting the business. What should be included in the deployment plan?

   • AA plan to grant all data scientists direct access to the production model for quick fixes.
   • BA plan to monitor the model's forecast accuracy and latency, with a defined process for updating the model based on performance.
   • CA plan to freeze the model's code and data so it never changes after deployment.
   • DA plan to delete the validation data after deployment to save storage costs.
   Reveal Solution Discussion Select an answer first
5. 34foundation · easy

   Which statistical method is commonly used to detect data drift by comparing the distribution of a feature in the current production data to the distribution in the training data?

   • ALinear regression
   • BKolmogorov-Smirnov (KS) test
   • CK-means clustering
   • DDecision tree classification
   Reveal Solution Discussion Select an answer first