bark: Bayesian Additive Regression Kernels

Bayesian Additive Regression Kernels (BARK) provides an implementation for non-parametric function estimation using Levy Random Field priors for functions that may be represented as a sum of additive multivariate kernels. Kernels are located at every data point as in Support Vector Machines, however, coefficients may be heavily shrunk to zero under the Cauchy process prior, or even, set to zero. The number of active features is controlled by priors on precision parameters within the kernels, permitting feature selection. For more details see Ouyang, Z (2008) "Bayesian Additive Regression Kernels", Duke University. PhD dissertation, Chapter 3 and Wolpert, R. L, Clyde, M.A, and Tu, C. (2011) "Stochastic Expansions with Continuous Dictionaries Levy Adaptive Regression Kernels, Annals of Statistics Vol (39) pages 1916-1962 <doi:10.1214/11-AOS889>.

Package details

AuthorMerlise Clyde [aut, cre, ths] (ORCID=0000-0002-3595-1872), Zhi Ouyang [aut], Robert Wolpert [ctb, ths]
MaintainerMerlise Clyde <>
LicenseGPL (>= 3)
Package repositoryView on CRAN
Installation Install the latest version of this package by entering the following in R:

Try the bark package in your browser

Any scripts or data that you put into this service are public.

bark documentation built on April 19, 2023, 1:10 a.m.