AWS Certified Machine Learning Engineer Associate · Free Practice Question Medium
Question 3
A machine learning engineer is responsible for deploying a model in Amazon SageMaker. After deployment, the engineer needs to monitor the model to detect data drift, model quality drift, and feature attribution drift over time. What are the most appropriate steps to take to monitor the deployed model effectively?
-
A
Implement a baseline job, define constraints based on training data, and schedule monitoring jobs.
-
B
Monitor only the model predictions for accuracy and retrain the model when performance declines.
-
C
Use Amazon S3 to store logs manually and monitor changes in the input data via manual inspection.
-
D
Perform model re-training without any active monitoring to avoid unnecessary processing costs.
Reveal correct answer
Correct answer: A
A.
SageMaker model monitoring allows the engineer to set up a baseline job to define constraints from the training data, followed by scheduling monitoring jobs to track data drift, model quality drift, and feature attribution drift automatically.
B.
While monitoring model predictions is important, this option does not mention setting up monitoring jobs or detecting various types of drift, such as data or feature attribution drift.
C.
Manually inspecting logs is not efficient for continuous monitoring. SageMaker model monitoring provides automated detection and alerts.
D.
Active monitoring is critical to detecting drift early and avoiding model performance degradation. Simply retraining without monitoring can lead to undetected issues in production.
Discussion
Think the marked answer is wrong, or have a better explanation? Share it below — comments appear after review.
