Using AWS Glue, an engineer must transform multiple similar time-series files from a vendor into the format required by the SageMaker DeepAR forecasting algorithm and compress them to reduce storage cost. Which approach satisfies these requirements?

The SageMaker DeepAR algorithm natively supports Apache Parquet as an input format, making it a suitable choice for data transformation. Parquet is a columnar storage format that is highly efficient for analytical queries and is widely used in big data ecosystems. XZ is a compression algorithm known for its high compression ratio, which directly addresses the requirement to reduce storage costs. While DeepAR can also accept RecordIO-Protobuf, Parquet is generally preferred for its broader ecosystem compatibility and analytical advantages. Snappy and gzip are faster compression algorithms but typically offer lower compression ratios than XZ, meaning they would be less effective at reducing storage costs.