MLOps Best Practices MACHINE-LEARNING-ASSOCIATE Practice Questions (Page 4)

Questions 16–20

1. 16application · medium

   A bank is defining its MLOps strategy for a fraud detection model. The model must be retrained frequently due to changing fraud patterns, but the team also needs to maintain audit trails for regulatory compliance. Which strategy best balances these needs?

   • AUse a single model version for all predictions to simplify compliance.
   • BAutomate retraining with MLflow tracking and log all model versions and data used.
   • CRetrain manually every quarter to keep the audit trail simple.
   • DAutomate retraining but do not log any metadata to save storage.
   Reveal Solution Discussion Select an answer first
2. 17foundation · easy

   Which practice is essential for ensuring reproducibility in an MLOps workflow?

   • AManually recording training results in a spreadsheet
   • BUsing a fixed random seed for all model training runs
   • CStoring all models in a single shared folder
   • DVersioning code, data, and model artifacts together
   Reveal Solution Discussion Select an answer first
4. 18expert · hard

   A financial company has a credit risk model that is retrained quarterly. The model is used for loan approvals, and the team wants to ensure that the model's decisions are explainable to regulators. They have implemented MLflow for tracking. What is the most important practice to support explainability?

   • ALog the model's feature importance and decision rules for each version in MLflow.
   • BRely on the data science team to explain decisions verbally.
   • CStore only the model artifact without any metadata.
   • DUse a complex model that is hard to interpret but has high accuracy.
   Reveal Solution Discussion Select an answer first
5. 19expert · hard

   A large e-commerce company has a recommendation model that is retrained daily. The team is experiencing frequent model performance issues in production, and they suspect it is due to data drift. They want to implement monitoring but have limited engineering resources. Which approach is most effective?

   • AImplement automated data drift detection on key features and alert the team when drift exceeds a threshold.
   • BRetrain the model more frequently without any monitoring.
   • CMonitor only the model's accuracy on a static test set.
   • DManually review feature distributions every week.
   Reveal Solution Discussion Select an answer first
6. 20foundation · easy

   What is a key best practice for monitoring machine learning models in production?

   • ATracking data drift and model performance over time
   • BMonitoring only the model's accuracy on the original test set
   • CRetraining the model daily regardless of performance
   • DDisabling monitoring after the model is deployed successfully
   Reveal Solution Discussion Select an answer first