Model retraining with the metadata store

In the previous lesson, you learned about the critical role that the metadata store plays in facilitating the full automation of MLOps pipelines. It enables automatic monitoring of prediction service functionality, and logs evaluation metrics that help detect performance decay over time. You also learned about triggered retraining and the importance of updating models to account for drift.

In this exercise, you will apply your knowledge by ordering a series of steps that demonstrate how the metadata store and triggered retraining can be used to update models automatically and maintain optimal performance.