An ML engineer needs to use data with Amazon SageMaker Canvas to train an ML model. The data is stored in Amazon S3 and is complex in structure. The ML engineer must use a file format that minimizes p

Question

Options

• ACSV files compressed with Snappy
• BJSON objects in JSONL format
• CJSON files compressed with gzip
• DApache Parquet files

• A
  13% (4)
• B
  10% (3)
• C
  3% (1)
• D
  73% (22)

Explanation

Apache Parquet is optimized for complex, structured data because it stores data in a columnar format, allowing SageMaker Canvas to read only the columns needed rather than scanning entire rows - dramatically reducing processing time for large, complex datasets. CSV (A) is row-based and uncompressed Snappy adds overhead without the structural benefits needed for complex data, making it slow to process. JSONL (B) is row-oriented and verbose, requiring full parsing of every record even when only a subset of fields is needed. Gzip-compressed JSON (C) must be fully decompressed before any data can be read, creating a sequential bottleneck that negates compression savings for complex datasets. Memory tip: Think "Parquet = Partial reads" - the columnar format lets the engine skip irrelevant data, which is the key to minimizing processing time in any analytics or ML context.

No community discussion yet for this question.