You received a CSV dataset in S3 that you will use for ML training. Before training, you must find missing or invalid values and count outliers with the least operational effort. Which approach meets this requirement?

The correct approach is to keep the data as CSV, import it into SageMaker Data Wrangler, and use the Data Quality and Insights report. SageMaker Data Wrangler is specifically designed for data preparation and analysis, offering built-in tools to identify missing/invalid values and detect outliers with minimal operational effort. Its Data Quality and Insights report automates these tasks, providing a comprehensive overview without requiring manual SQL queries or custom code. Converting to Parquet with AWS Glue and querying with Athena (options 1 and 3) adds unnecessary steps and operational overhead for this specific task. While Parquet is efficient for storage and querying large datasets, it's not required for initial data profiling within Data Wrangler. Running an AWS Glue crawler and querying with Athena (options 1 and 2) would require writing and executing SQL queries to find missing values and outliers, which is more effort than Data Wrangler's automated reports.