AIF-C01 Question 51
Select 4You are building a machine learning pipeline on AWS to predict customer churn. The pipeline includes data preprocessing, feature engineering, model training, and monitoring the deployed model. Which combination of AWS services should you use to ensure an efficient and scalable implementation?
- A
Amazon SageMaker Data Wrangler for preprocessing and feature engineering
- B
Amazon SageMaker Feature Store for storing and serving engineered features
- C
AWS CloudTrail for training the machine learning model
- D
Amazon SageMaker for training and deploying the model
- E
Amazon SageMaker Model Monitor for monitoring the deployed model
Show answer and explanation
Correct answers: A, B, D, E
Explanation
To build a complete ML pipeline on AWS, several specialized services can be used for different stages. Amazon SageMaker Data Wrangler handles preprocessing and feature engineering. Amazon SageMaker Feature Store manages feature storage and reuse. Amazon SageMaker is used for model training and deployment. Finally, Amazon SageMaker Model Monitor ensures that the deployed model performs as expected by monitoring for data drift and other anomalies. AWS CloudTrail, while valuable for auditing and logging, does not play a role in the ML pipeline.
- A. Correct.
Amazon SageMaker Data Wrangler simplifies the data preprocessing and feature engineering process by allowing users to clean, transform, and visualize data without needing to write extensive code. It is suitable for the preprocessing stage of the pipeline.
- B. Correct.
Amazon SageMaker Feature Store is a fully managed repository to store and retrieve machine learning features. It is designed for efficient feature reuse, enabling consistency across training and inference.
- C. Incorrect.
AWS CloudTrail is a service for monitoring and logging API activity across AWS accounts. While useful for security and compliance, it is not used for training machine learning models.
- D. Correct.
Amazon SageMaker provides a fully managed service for training and deploying machine learning models at scale. It is the appropriate choice for the training and deployment stages of the pipeline.
- E. Correct.
Amazon SageMaker Model Monitor continuously monitors the deployed model to detect data drift and other issues, making it ideal for the monitoring stage of the pipeline.