Abstract
In this paper, the new Bayesian approach based on normal scale mixture is proposed for estimating the quantile regression of the single index model and selecting variables. The Gaussian process prior have been considered to the nonparametric link function. Bayesian hierarchical model construct and Gibbs sampler algorithm using for posterior inference. Simulation study was addressed to compare our suggested method with three other existing methods. The simulation studies significant that the new suggested method offers substantial improvement over the other three methods..
