MLOps Best Practices MACHINE-LEARNING-ASSOCIATE Practice Questions (Page 5)

Questions 21–22

1. 21application · medium

   A media company uses a content recommendation model that is updated daily. The data science team wants to ensure that the model in production is always the one that passed validation. What is the best way to achieve this?

   • AManually promote models to production after a quick review.
   • BUse the same model for a month before updating.
   • CAutomate the promotion of a model from Staging to Production only after validation tests pass.
   • DDeploy all new models directly to production and monitor performance.
   Reveal Solution Discussion Select an answer first
2. 22expert · hard

   A healthcare company is deploying a model that predicts patient readmission. The model is retrained monthly, and the team wants to ensure that the model remains fair across different patient demographics. They have limited resources for monitoring. What is the most effective way to monitor for fairness?

   • AMonitor only overall accuracy and ignore demographic differences.
   • BAutomatically compute fairness metrics across demographic groups after each retraining and alert if they degrade.
   • CManually review fairness metrics once a year.
   • DRely on the data science team to ensure fairness without any formal monitoring.
   Reveal Solution Discussion Select an answer first