Pandas is a Python library that provides flexible and powerful data structures and tools for data analysis and manipulation. The library centers on two primary data structures: Series for one-dimensional labeled data and DataFrame for two-dimensional labeled data, both designed to handle relational and labeled data in ways similar to R's data.frame objects. The project originated at AQR, a quantitative hedge fund, in 2008 and has been under continuous active development since then.
The library excels at handling missing data represented as NaN, NA, or NaT across both floating-point and non-floating-point data types. Pandas provides robust capabilities for data alignment, allowing Series and DataFrame objects to automatically align data during computations or to be explicitly aligned to specific label sets. The library includes powerful group by functionality for split-apply-combine operations, enabling both data aggregation and transformation. Users can perform intelligent label-based slicing, fancy indexing, and subsetting of large datasets, along with flexible merging and joining of multiple datasets.
Pandas offers comprehensive reshaping and pivoting capabilities, hierarchical axis labeling with multiple labels per tick, and extensive I/O tools for loading data from flat files like CSV and delimited formats, Excel files, databases, and the ultrafast HDF5 format. The library includes specialized time series functionality such as date range generation, frequency conversion, moving window statistics, and date shifting and lagging operations. Size mutability allows columns to be inserted and deleted from DataFrames and higher-dimensional objects.
The repository maintains active development with substantial community engagement.
Pandas depends on NumPy for large multi-dimensional arrays and mathematical operations, python-dateutil for datetime extensions, and tzdata for IANA time zone database support on Windows and Emscripten platforms. The library is distributed under the BSD 3-Clause license and is powered by NumFOCUS. Installation is available through PyPI and Conda, with source installation requiring Cython in addition to standard dependencies. The project maintains comprehensive documentation on PyData.org and fosters community engagement through GitHub issue tracking, a dedicated mailing list, Slack channels, and regular community meetings including monthly sessions for new contributors.
