Certified Machine Learning ProfessionalDemo

Question

A Machine Learning Engineer maintains a fraud-detection model deployed on Databricks. They need to implement a retraining pipeline that automatically begins when the model’s F1 score falls below a threshold or when input-feature distributions change substantially.

Which two actions should the engineer take to implement this automated retraining?

Choose two

A Schedule a recurring query on the Lakehouse monitoring table.
B Use MLflow to manually log metrics and retrain the model offline.
C Set up a manual retraining schedule to run every week regardless of alerts.
D Configure these alerts to send webhook notifications that trigger the model training job.
E Use Databricks SQL to create alerts on model performance and data drift metrics stored in Delta tables.

Explanation

Databricks Lakehouse Monitoring stores model-performance and distribution-drift metrics in Delta tables that can be queried with Databricks SQL. Databricks SQL alerts can evaluate threshold conditions on those metrics and notify configured destinations when the condition is met. Configuring webhook notifications enables the alert event to invoke the model-training job, providing retraining based on F1 degradation or feature drift rather than a fixed schedule.

Learn more

Certified Machine Learning ProfessionalDemo

AstroTutor

Notes

Readiness

QuestionQ141

Community Discussion