Objective 4.3 Maintain the Model Postproduction AIP-210 Practice Questions (Page 1)

Questions 1–5

1. 1expert · hard

   A government agency uses a model to allocate resources for disaster response. The model was developed by a contractor who has since gone out of business. The agency needs to maintain the model internally but has limited documentation. The model is critical and must remain operational. What is the BEST first step for the agency?

   • AHire a new contractor to rebuild the model from scratch.
   • BKeep the model running without documentation and hope it continues to work.
   • CRetire the model immediately and develop a new one from scratch.
   • DCreate a model documentation package by reverse-engineering the model and interviewing any available stakeholders.
   Reveal Solution Discussion Select an answer first
2. 2foundation · easy

   What is the primary criterion for retiring a model from production?

   • AA newer model version is available
   • BThe model has been in production for a fixed number of days
   • CThe model's training data has been deleted
   • DThe model is no longer effective or needed for its intended purpose
   Reveal Solution Discussion Select an answer first
4. 3foundation · easy

   Which practice is essential for effective model versioning?

   • ARecording the training dataset, hyperparameters, and evaluation metrics for each version
   • BAssigning the same version number to all models in a project
   • CStoring only the final model artifact without any metadata
   • DDeleting old versions to save storage space
   Reveal Solution Discussion Select an answer first
5. 4foundation · easy

   What is the primary benefit of versioning machine learning models?

   • AIt allows tracking of changes and enables rollback to a previous version if needed
   • BIt automatically improves the model's accuracy with each version
   • CIt reduces the need for model documentation
   • DIt ensures the model is deployed to multiple environments simultaneously
   Reveal Solution Discussion Select an answer first
6. 5expert · hard

   A telecommunications company operates a customer churn model. Monitoring shows that the model's AUC has dropped from 0.85 to 0.78 over the past month. The data science team investigates and finds that the distribution of customer tenure has shifted significantly because of a new acquisition campaign. The model's training data did not include such short-tenure customers. The team needs to restore performance quickly while minimizing risk. What should they do?

   • AAdjust the decision threshold to increase the model's sensitivity.
   • BRoll back to the previous model version that had higher AUC.
   • CRetrain the model using a mix of historical data and recent data that includes the new customer segment.
   • DFreeze the model and wait for the next scheduled retraining in three months.
   Reveal Solution Discussion Select an answer first