SkdP

SkdP is a software library for nonparametric density estimation that applies skewed kernel methods to better capture asymmetry in data distributions. The project aims to provide an end-to-end workflow for estimating, validating, and visualizing probability densities within a single framework. In developer documentation, SkdP is typically expanded as Skewed Kernel Density Processor.

Key features include multivariate skewed kernel density estimators, adaptive bandwidth selection, cross-validation for model tuning, and tools for evaluating fit. The library offers Python bindings with optional R wrappers, a command-line interface, and support for common data formats such as CSV and Parquet. It emphasizes interoperability with NumPy, SciPy, and plotting libraries.

Architecture centers on a core density estimator module, a visualization layer, and a utilities toolkit for preprocessing and diagnostics. It supports datasets with thousands of observations and can leverage optional GPU acceleration for larger problems. Typical workflow involves data preprocessing, model fitting, density evaluation, and visualization of marginal and conditional densities.

History and reception: SkdP originated from a collaboration among statistics researchers and open-source developers in the mid-2010s. The project has had several major releases, with ongoing community maintenance and contributions. It is used in academic courses and industry analytics pipelines, and is cited for providing flexible density estimation without relying on restrictive parametric assumptions. Some users note higher memory usage compared with traditional parametric methods.

See also: kernel density estimation, nonparametric statistics.