Bayesian Regression with Errors from ESDIW Distribution

• Activity Number: 38
• Type: Contributed
• Date/Time: Sunday, August 4, 2013 : 2:00 PM to 3:50 PM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Ahmad Flaih*+ and Jose Guardiola and Hassan Elsalloukh
• Companies: Al-Qadisiya University and Texas A&M University-CC and University of Arkansas at Little Rock
• Keywords: Bayesian Regression ; Informative Prior ; Noninformative Prior ; Jeffrey's Rule ; Nuisance Parameter

Abstract

In this paper our goal is to make statistical inference on using Bayesian techniques for deriving the posterior density function of the regression coefficient vector based on Informative and Noninformative prior distribution. We assume that the error term in the proposed model distributed according to Epsilon Skew Double Inverted Weibull (ESDIW) distribution. In the Informative analysis, the interest parameter is the regression coefficient parameter , so the prior distribution of  that represent our beliefs assumed to be ESDIW, whereas in the noninformative analysis, we derive the independent Jeffrey's priors based on Jeffrey's General rule and certain groups of the interest and nuisance parameters.