CDFsources

CDFsources is a software library that provides tools and data sources for constructing and analyzing cumulative distribution functions (CDFs) from data. It emphasizes reproducibility, compatibility with common data formats, and a modular design that supports empirical and parametric approaches to CDF estimation. The project aims to standardize workflows for deriving CDFs across disciplines such as statistics, finance, engineering, and data science.

Key features include ingestion of data in CSV, JSON, or database inputs; estimation methods such as empirical CDF, kernel-smoothed CDF, and fits to common distributions; support for confidence intervals via bootstrap or asymptotic methods; visualization capabilities for plotting CDFs and comparison across groups; and export options in vector formats (SVG, PDF) and programmatic access through language bindings.

Architecturally, CDFsources uses a modular core with separate estimator, I/O, and visualization components. It supports a plug-in architecture enabling new estimators and distribution families to be added without altering core code. It offers APIs for Python and R, a command-line interface, and documentation to guide integration into data pipelines and notebooks.

Typical use cases include assessing distributional assumptions in hypothesis testing, estimating risk measures that depend on tail behavior, quality control through CDF-based thresholds, and teaching statistics by illustrating empirical results. The project is maintained by an open-source community and released under a permissive license.

Related topics include cumulative distribution function, empirical distribution function, kernel density estimation, and bootstrap methods.