English

Bayesian Variable Selection for Linear Regression with the $\kappa$-$G$ Priors

Methodology 2016-10-20 v2

Abstract

In this paper, we introduce a new methodology for Bayesian variable selection in linear regression that is independent of the traditional indicator method. A diagonal matrix G\mathbf{G} is introduced to the prior of the coefficient vector β\boldsymbol{\beta}, with each of the gjg_j's, bounded between 00 and 11, on the diagonal serves as a stabilizer of the corresponding βj\beta_j. Mathematically, a promising variable has a gjg_j value that is close to 00, whereas the value of gjg_j corresponding to an unpromising variable is close to 11. This property is proven in this paper under orthogonality together with other asymptotic properties. Computationally, the sample path of each gjg_j is obtained through Metropolis-within-Gibbs sampling method. Also, in this paper we give two simulations to verify the capability of this methodology in variable selection.

Keywords

Cite

@article{arxiv.1503.06370,
  title  = {Bayesian Variable Selection for Linear Regression with the $\kappa$-$G$ Priors},
  author = {Zichen Ma and Ernest Fokoué},
  journal= {arXiv preprint arXiv:1503.06370},
  year   = {2016}
}

Comments

19 pages, 3 figures, 1 table, 1 algorithm

R2 v1 2026-06-22T08:58:48.960Z