MLA-C01 Question 263
Select 2A data science team has deployed a machine learning model to predict customer churn. The team wants to implement two different serving methods: one for real-time predictions during customer interactions and another for generating churn probability reports for all customers at the end of each day. Which combination of services and configurations would best meet these requirements?
- A
Use Amazon SageMaker Hosting Services with an endpoint for real-time inference and AWS Glue for batch processing.
- B
Use Amazon SageMaker Batch Transform for both real-time and batch inference.
- C
Use Amazon SageMaker Hosting Services with an endpoint for real-time inference and Amazon SageMaker Batch Transform for batch inference.
- D
Use AWS Lambda for real-time inference and Amazon SageMaker Processing Jobs for batch inference.
- E
Use Amazon SageMaker Hosting Services for real-time inference and Amazon Athena to query batch inference results stored in Amazon S3.
Show answer and explanation
Correct answers: A, C
Explanation
The correct approach to serve ML models depends on the use case. For real-time inference, Amazon SageMaker Hosting Services provides low-latency endpoints optimized for handling real-time requests. For batch inference, Amazon SageMaker Batch Transform or AWS Glue can be used to process large datasets and generate predictions at scale. Combining these services ensures that both real-time and batch inference requirements are met efficiently.
- A. Correct.
This is a correct answer. Amazon SageMaker Hosting Services is well-suited for real-time inference as it provides low latency endpoints. AWS Glue can be used for batch processing by running ETL jobs that process the data and send it for inference using a model hosted in SageMaker.
- B. Incorrect.
This is incorrect. Amazon SageMaker Batch Transform is specifically designed for batch inference and is not suitable for real-time inference as it does not provide low-latency predictions.
- C. Correct.
This is a correct answer. Amazon SageMaker Hosting Services provides real-time endpoints with low latency, while SageMaker Batch Transform is specifically designed for batch inference, making this a suitable combination.
- D. Incorrect.
This is incorrect. AWS Lambda can handle real-time inference but is not ideal for long-running or complex batch jobs. SageMaker Processing Jobs are better suited for data preprocessing or feature engineering but not for batch inference.
- E. Incorrect.
This is incorrect. While Amazon SageMaker Hosting Services can handle real-time inference, Amazon Athena is a query service and is not designed to run batch inference jobs. Athena is more suitable for querying results already stored in Amazon S3.