AS 308

Algorithm AS 308: Highest posterior density credible region and minimum area confidence region: The bivariate case. The authors [J. Stat. Comput. Simul. 44, 243-250 (1993)] presented an algorithm, called HPD, to compute the highest posterior density (HPD) region in the univariate case. Depending on the nature of the distribution considered, the 100(1-α)% credible region can be an interval or a set of disjoint intervals. This algorithm has been successfully used in several science and engineering applications. This algorithm is hereby extended to cover the bivariate case, continuous and discrete, and is called BHPD. At the same time for the continuous case it can be used to compute the 100(1-α)% confidence region with minimum volume in classical statistics, under suitable conditions imposed on the pivotal quantity.