Welcome to cuDF’s documentation!

cuDF is a Python GPU DataFrame library (built on the Apache Arrow columnar memory format) for loading, joining, aggregating, filtering, and otherwise manipulating data. cuDF also provides a pandas-like API that will be familiar to data engineers & data scientists, so they can use it to easily accelerate their workflows without going into the details of CUDA programming.

Contents:

• User Guide
  • 10 Minutes to cuDF and Dask-cuDF
  • Comparison of cuDF and Pandas
  • Supported Data Types
  • Input / Output
  • Working with missing data
  • GroupBy
  • Overview of User Defined Functions with cuDF
  • Interoperability between cuDF and CuPy
  • Options
  • Performance comparisons
  • Pandas Compatibility Notes
  • Copy-on-write
• API reference
  • Series
  • DataFrame
  • Index objects
  • GroupBy
  • General Functions
  • General Utilities
  • Window
  • Input/output
  • SubwordTokenizer
  • String handling
  • List handling
  • Struct handling
  • Options and settings
• Developer Guide
  • Library Design
  • Contributing Guide
  • Writing documentation
  • Testing cuDF
  • Benchmarking cuDF
  • Options
  • pylibcudf

Indices and tables

• Index

• Search Page