PDF

Keywords

Composite quantile regression
new hierarchical algorithm
new Gibbs sampler
new regularization approach

Abstract

ــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــAbstract: In this paper, we propose an efficient and easy method for achieving the coefficients estimatedand variable selection in composite quantile regression via the employed new regularization method. It is the utilization of a new prior distribution that consists of a mixture of the normal distribution with the exponential distribution as an alternative form to the Laplace distribution. This composition will provide us with a good and efficient algorithm that is both fast and stable in estimating model parameters. The simulation scenario and real data were used in this study.
  PDF