MLS-C01 Question 361
Select 3You are a Machine Learning Specialist deploying a machine learning model for inference in a production environment. The model is hosted on Amazon SageMaker, and you need to expose an endpoint to allow your application to send requests for real-time predictions. Additionally, you want to monitor the endpoint's performance and capture input data for further analysis. Which of the following steps should you take to achieve this?
- A
Deploy the model on an Amazon SageMaker endpoint and enable the Data Capture configuration.
- B
Use Amazon API Gateway to create an API for the SageMaker endpoint.
- C
Enable CloudWatch Logs and Metrics for the SageMaker endpoint.
- D
Configure the endpoint to return predictions in JSON format.
- E
Set up an Amazon S3 bucket to store captured data for analysis.
Show answer and explanation
Correct answers: A, C, E
Explanation
To expose an endpoint and interact with it in Amazon SageMaker, you deploy your model to a SageMaker endpoint. Enabling Data Capture allows you to log input and output data for further analysis. Monitoring the performance of the endpoint through CloudWatch Logs and Metrics is critical to ensure its optimal operation. Finally, an S3 bucket is necessary to store the captured data. Using Amazon API Gateway or explicitly configuring JSON output is not strictly required for this workflow.
- A. Correct.
Correct: Deploying the model on an Amazon SageMaker endpoint is necessary to expose it for real-time inference, and enabling Data Capture allows you to collect input data for further analysis.
- B. Incorrect.
Incorrect: While Amazon API Gateway can be used to create APIs, it is not required to expose a SageMaker endpoint, as SageMaker endpoints are already accessible via RESTful APIs.
- C. Correct.
Correct: Enabling CloudWatch Logs and Metrics allows you to monitor the performance of the SageMaker endpoint, such as latency and invocation metrics.
- D. Incorrect.
Incorrect: Returning predictions in JSON format is the default behavior of SageMaker endpoints and does not require explicit configuration.
- E. Correct.
Correct: An Amazon S3 bucket is needed to store the captured data for analysis when using SageMaker's Data Capture feature.