KernelMix
Copyright (C) 2026 Nilson Chapagain

KernelMix implements selected Bayesian kernel machine regression methods described
by Bobb et al. (2015), DOI 10.1093/biostatistics/kxu058, and Bobb et al. (2018),
DOI 10.1186/s12940-018-0413-y.

The R package bkmr 0.2.2, by Jennifer F. Bobb and contributors, is a reference for
prior conventions, likelihoods and posterior summaries. Its GPL-2 source informed
this implementation. The installed package is used in optional reference checks.
KernelMix uses distinct Python code and sampling proposals. Count/survival
fitting uses elliptical slice sampling (Murray, Adams and MacKay, 2010,
https://proceedings.mlr.press/v9/murray10a.html). Standard count and survival
likelihood references and selected priors are identified in docs/methods.md.
The original authors have not endorsed this package. Source papers and their
figures are not bundled.

KernelMix code and documentation are distributed under GNU GPL version 2 only,
without warranty. See LICENSE. NumPy, SciPy and Matplotlib retain their own licenses.

Additional kernel definitions follow Rasmussen and Williams (2006), Gaussian Processes for Machine Learning, chapter 4. Random Fourier features follow Rahimi and Recht (2007). Predictive processes follow Banerjee et al. (2008), DOI 10.1111/j.1467-9868.2008.00663.x. Residual variance/correlation and survival data conventions are separately cited in the structured-model guide. These are model ingredients, not claims of new estimators or complete upstream interfaces.

The bundled airquality CSV is exported from R 4.5.0 datasets::airquality. The original measurements are credited to the New York State Department of Conservation and the National Weather Service; see Chambers et al. (1983), Graphical Methods for Data Analysis, and the R datasets documentation. This export and its test fixture are distributed under GPL-2.0 as part of this package.
