ai-300 Practice Question 2017
Exam: ai-300
Domain: Monitor and manage ML deployments
Topic: Monitor model performance and data drift
Difficulty: medium
In an Azure Machine Learning deployment, you notice that the performance of your model has degraded over time. You want to monitor for data drift to understand if changes in the input data have caused this degradation. Which Azure service should you use to set up a data drift monitor and how should you configure it?
Answer Options
A
Use Azure Monitor with custom logs
B
Set up an Azure Machine Learning Data Drift Monitor with a specified baseline dataset and drift check frequency
C
Deploy a custom script using Azure Functions to periodically check for drift
D
Configure an Azure DevOps pipeline to run model validation tests
Correct Answer
B: Set up an Azure Machine Learning Data Drift Monitor with a specified baseline dataset and drift check frequency
Explanation
The correct answer is to use the Azure Machine Learning Data Drift Monitor. It allows you to specify the baseline dataset and the frequency at which the drift should be checked. The other options are incorrect because they do not provide the specific functionality required for monitoring data drift.