Elastic Net Regression with the Value of the L2 Penalty Parameter Associated with Bayesian Analysis

Authors

  • Kanyalin Jiratchayut Department of Mathematics and Statistics, Faculty of Science and Technology, Thammasat University
  • Chinnaphong Bumrungsup Department of Mathematics and Statistics, Faculty of Science and Technology, Thammasat University

Keywords:

Bayes factor, Bayesian analysis, elastic net, L2 penalty, shrinkage

Abstract

The aim of this article is to propose the method for choosing the value of the L2 penalty parameter, \lambda_{2}, of elastic net linear regression model using Bayesian analysis. The value of \lambda_{2} is specified through the behavior of Bayes factor. We study the performance of elastic net estimators where the value of  is based on Bayes factor and the value of \lambda_{2}  is chosen by 10-fold cross-validation method. Simulation studies and real data examples show that the elastic net estimator where the value of  \lambda_{2} is based on Bayes factor performs better in prediction accuracy.

Downloads

Published

2015-07-30

How to Cite

Jiratchayut, K., & Bumrungsup, C. (2015). Elastic Net Regression with the Value of the L2 Penalty Parameter Associated with Bayesian Analysis. Thailand Statistician, 13(2), 243–269. Retrieved from https://ph02.tci-thaijo.org/index.php/thaistat/article/view/37775

Issue

Vol. 13 No. 2 (2015): July

Section

Articles