Bayesian Analysis of Skewed Regression Model in the “Large P, Small N” Paradigm

This paper performed the skewed regression model whose data set of n observed responses is to be related to a set of p potential predictors with p >> n. We introduce the skewed regression models based on the ideas proposed by Chen, Dey and Shao(1999). We use singular-value decompositions of matrices which have measured values of large numbers of predictors across samples, generating factor representations and possibly massive dimension reduction. Also, we develop easily implemented and standard Markov chain Monte Carlo(MCMC) methods to produce posterior inferences on the high-dimensional regression parameter. Finally, we have the simulation study.