MLA-C01 Question 383
Select 3You are designing a machine learning system on AWS to predict customer churn. As part of the design process, you need to implement monitoring principles to ensure the system performs reliably over time. Which monitoring practices align with the AWS Well-Architected Machine Learning Lens design principles?
- A
Set up Amazon CloudWatch to monitor input data quality and distribution over time.
- B
Deploy an offline batch inference system that does not require monitoring since it processes data in isolation.
- C
Implement Amazon SageMaker Model Monitor to detect changes in model predictions and data drift.
- D
Use automated alarms in Amazon CloudWatch to alert stakeholders when model latency exceeds a threshold.
- E
Rely on periodic manual reviews of system logs to check for anomalies in model behavior.
Show answer and explanation
Correct answers: A, C, D
Explanation
The AWS Well-Architected Machine Learning Lens emphasizes continuous monitoring of data quality, model performance, and system behavior to ensure reliable and accurate predictions over time. Tools like Amazon CloudWatch and Amazon SageMaker Model Monitor enable proactive and automated monitoring, while manual efforts or lack of monitoring fail to meet these best practices.
- A. Correct.
Monitoring input data quality and distribution over time is critical to detecting data drift or quality issues that can impact model performance. Amazon CloudWatch is a suitable tool for this purpose.
- B. Incorrect.
Offline batch inference systems still require monitoring to ensure reliability and detect issues like data drift or model degradation. Not monitoring them would violate Well-Architected principles.
- C. Correct.
Amazon SageMaker Model Monitor is specifically designed to detect changes in data quality, data drift, and model performance, aligning with Well-Architected Machine Learning Lens principles for monitoring.
- D. Correct.
Automated alarms in Amazon CloudWatch help stakeholders respond promptly to issues like increased latency, ensuring system performance and reliability.
- E. Incorrect.
Manual reviews are not scalable or proactive and do not align with the Well-Architected Machine Learning Lens, which emphasizes automation and continuous monitoring.