Objective 4.3 Maintain the Model Postproduction AIP-210 Practice Questions (Page 5)

Questions 21–25

1. 21application · medium

   A data science team is developing a churn prediction model. The team is small, and the lead data scientist is leaving the company. Management is concerned about knowledge transfer and future maintenance. What should the team do to ensure the model can be maintained after the lead leaves?

   • AAsk the lead to record a video walkthrough of the code and share it with the team.
   • BCreate comprehensive documentation covering data sources, feature engineering, model selection, and deployment steps.
   • CHire a consultant to rebuild the model from scratch after the lead leaves.
   • DKeep the model unchanged until the new hire is trained.
   Reveal Solution Discussion Select an answer first
2. 22expert · hard

   A software company maintains a model that predicts user engagement. The model is updated every two weeks. The product team often asks for the predictions from a specific model version to compare with current behavior. The data science team needs to ensure that any model version can be easily retrieved and used for analysis. What should they implement?

   • AA shared folder with model files named by date.
   • BA documentation wiki that describes each model version.
   • CA database that stores the model's predictions for each version.
   • DA model registry with versioning and the ability to load any specific version on demand.
   Reveal Solution Discussion Select an answer first
4. 23expert · hard

   A financial services firm has a model that predicts loan default. The model has been in production for five years. Recently, the firm's risk committee decided to replace it with a new model that uses alternative data. The old model is still referenced in several regulatory reports. The firm must retire the old model but ensure that the reports can still be produced. What should the firm do?

   • AKeep the old model running solely for report generation until the reports are no longer required.
   • BRetire the model but archive it and all its metadata so that the reports can be regenerated if needed.
   • CRetrain the old model with recent data to keep it accurate for reporting.
   • DDelete the old model and its data to avoid confusion with the new model.
   Reveal Solution Discussion Select an answer first
5. 24application · medium

   A retail chain's inventory forecasting model has been in production for three years. The company has shifted from physical stores to an online-only model, and the model's forecasts are no longer used by the operations team. The model still runs daily and incurs compute costs. What should the team do?

   • AReduce the model's execution frequency to weekly to save costs.
   • BRetrain the model with online sales data to keep it relevant.
   • CRetire the model, document the reason, and archive its artifacts for potential future use.
   • DKeep the model running in case the company returns to physical stores.
   Reveal Solution Discussion Select an answer first
6. 25foundation · easy

   A data science team has deployed a classification model to production. Which activity is the primary purpose of ongoing model monitoring?

   • ATo detect degradation or drift in model performance over time
   • BTo replace the model with a newer version immediately upon any metric change
   • CTo automatically retrain the model whenever new data arrives
   • DTo ensure the model is deployed on the most cost-effective infrastructure
   Reveal Solution Discussion Select an answer first