arrow gives R users access to the Apache Arrow C++ library, a language-independent columnar memory format designed for efficient analytics. It provides dplyr-compatible operations on datasets too large to fit in memory, and enables zero-copy data sharing between R and Python.

What’s covered#

  • Arrow data structures – Table and Dataset objects
  • Read individual files – read_parquet(), read_feather(), read_csv_arrow()
  • Read multi-file datasets – open_dataset() with format and partitioning options
  • Write individual files – write_parquet(), write_feather(), compression options
  • Write multi-file datasets – write_dataset() with partitioning
  • Manipulate larger-than-memory datasets – dplyr verbs with collect()
  • Zero-copy R and Python data sharing – reticulate and pyarrow integration
  • Cloud storage support (S3) – reading and writing from S3 URIs