Bayesian lasso in quantile regression with a new prior
Fadel Hamid Hadi Alhusseini
Remah Oday Hassan
Al-Qadisiyah University – College of Administration and Economics
Corresponding Author: Remah Oday Hassan
Abstract : In this paper we propose new Bayesian hierarchical model for the lasso quantile regression
model based on new formulation for the laplace prior distribution of the interested parameter the
formulation use the scale mixture of uniform distribution mixing with standard exponential distribution .
Furthermore, new MCMC Gibbs sampler algorithm has proposed to estimate the parameters. Simulation
scenarios have conducted to study the behavior of the proposed model through the estimation accuracy and
variable selection procedure. Also, real data analysis illustrated that the proposed model outperforms some
other models.
Keywords: Bayesian hierarchical model, lasso, quantile regression Gibbs sample algorithm
