AIF-C01 Question 50
Select 3A machine learning team is building an ML pipeline for a retail company to predict customer churn. They need to: 1) preprocess and clean the raw data, 2) store and retrieve engineered features for model training, and 3) monitor the model in production for data drift. Which combination of AWS services is most appropriate to address these requirements?
- A
Amazon SageMaker Data Wrangler
- B
Amazon SageMaker Feature Store
- C
Amazon SageMaker Model Monitor
- D
Amazon Comprehend
- E
Amazon Rekognition
Show answer and explanation
Correct answers: A, B, C
Explanation
The appropriate combination of services for this ML pipeline includes Amazon SageMaker Data Wrangler for data preprocessing, Amazon SageMaker Feature Store for feature storage and retrieval, and Amazon SageMaker Model Monitor for monitoring the deployed model. These services address the key stages of preprocessing, feature management, and monitoring in the ML pipeline. Amazon Comprehend and Amazon Rekognition are not relevant to the problem since they are specific to NLP and computer vision tasks, respectively.
- A. Correct.
Amazon SageMaker Data Wrangler is used to preprocess and clean raw data, making it the appropriate service for data preparation in this scenario.
- B. Correct.
Amazon SageMaker Feature Store is designed to store and retrieve engineered features, which is essential for model training and reuse of features across teams.
- C. Correct.
Amazon SageMaker Model Monitor provides tools for monitoring models in production, such as detecting data drift or anomalies, which is critical for the monitoring stage of the ML pipeline.
- D. Incorrect.
Amazon Comprehend is a managed AI service for natural language processing tasks like sentiment analysis, which is not relevant to this ML pipeline scenario.
- E. Incorrect.
Amazon Rekognition is a managed AI service for computer vision tasks like image and video analysis, which is not relevant to this ML pipeline scenario.