MLA-C01 Domain 1: Data Preparation for Machine Learning

The DevOps team is automating releases and must split data into train/validation/test without leakage. Which design fits best? split data into train/validation/test without leakage.

Multiple choice — select one answer

Answer choices

  1. a. Oversampling, weights, or SMOTE techniques in SageMaker
  2. b. Amazon SageMaker Data Wrangler
  3. c. AWS Glue Data Catalog and crawlers
  4. d. SageMaker Processing or sklearn split with stratification

Architectural breakdown & explanation

Proper splits prevent optimistic metrics.

Correct answer(s): d. SageMaker Processing or sklearn split with stratification

Exam domain: Domain 1: Data Preparation for Machine Learning (28% weight on the official guide).

Official documentation & study links

Practice this certification

This page is one question from the AWS Certified Machine Learning Engineer – Associate bank. Run a full timed practice exam with domain-weighted random questions in your browser.