A data science team stores a tabular dataset in Amazon S3. The team wants to try various feature transformations (for example, categorical encoding), visualize the resulting distributions, and then automate the chosen feature-processing workflow. Which solution provides the most operational efficiency?

The correct answer leverages Amazon SageMaker Data Wrangler for its integrated capabilities. Data Wrangler allows data scientists to explore and apply preconfigured feature transformations, visualize the results directly using built-in templates, and then seamlessly export the entire processing flow as a SageMaker Pipeline. This provides the most operational efficiency because it keeps the entire workflow – exploration, transformation, visualization, and automation – within the SageMaker ecosystem, minimizing context switching and manual coding. The incorrect options introduce unnecessary complexity or break the workflow across multiple services. Using a SageMaker notebook instance requires manual coding for transformations and visualizations, and packaging into Lambda functions adds operational overhead. AWS Glue Studio with custom code and QuickSight also splits the workflow and increases development effort. While Data Wrangler can save transformations to S3, packaging each into individual Lambda functions and orchestrating with Step Functions is overly complex compared to a single SageMaker Pipeline export.