Question

A retail demand-forecasting model deployed on a Vertex AI endpoint has been serving stable predictions for six months. Model Monitoring suddenly fires drift alerts for three input features on the same day: 'store_region' (categorical), 'promo_flag' (categorical), and 'weekly_units_sold_lag1' (numeric). Business metrics tied to the model have not degraded yet. Before triggering an expensive retrain, what is the most appropriate first troubleshooting step?

Accepted Answer

Investigate the upstream data pipeline for a schema or encoding change, because simultaneous drift across unrelated features on the same day strongly suggests a data-source or ingestion issue rather than genuine distribution shift.

Answer

Immediately trigger the automated retraining pipeline, since three simultaneous feature drift alerts confirm concept drift.

Answer

Disable Model Monitoring drift alerting for these features to reduce alert noise while the model continues to serve.

Answer

Lower the drift alert thresholds so future shifts are caught earlier, then wait for ground-truth labels to arrive.