MLA-C01 Question 326
Select 3You are building a machine learning pipeline on AWS to predict customer churn using tabular data stored in Amazon S3. You plan to preprocess the data, train the model, and deploy it for inference. Which combination of AWS services can be used to efficiently handle this workflow?
- A
Amazon SageMaker Processing for data preprocessing
- B
Amazon Redshift ML for deploying the model
- C
Amazon SageMaker Training for model training
- D
AWS Lambda for model training
- E
Amazon SageMaker Endpoint for real-time inference
Show answer and explanation
Correct answers: A, C, E
Explanation
A complete machine learning workflow on AWS involves data preprocessing, model training, and deployment. Amazon SageMaker Processing efficiently handles preprocessing tasks, Amazon SageMaker Training is used for scalable model training, and Amazon SageMaker Endpoint is designed for real-time inference. Redshift ML and Lambda, while useful in other scenarios, do not fit the requirements for this pipeline.
- A. Correct.
Amazon SageMaker Processing is specifically designed for data preprocessing tasks, such as cleaning, transforming, and feature engineering within a machine learning pipeline.
- B. Incorrect.
Amazon Redshift ML is intended for integrating machine learning models directly into Redshift queries but is not suitable for deploying a model for real-time inference.
- C. Correct.
Amazon SageMaker Training is used for training machine learning models at scale, making it the appropriate service for this step in the workflow.
- D. Incorrect.
AWS Lambda can trigger machine learning workflows but is not designed for heavy computational tasks like model training.
- E. Correct.
Amazon SageMaker Endpoint is the correct service for deploying machine learning models to serve real-time inference.