Highest posterior mass prediction intervals for binomial and Poisson distributions (Q1785796)

The authors consider the problem of constructing prediction intervals (PIs) for the binomial and the Poisson distribution. They propose new highest posterior mass (HPM) PIs based on fiducial approach and compare them with an exact PI and an approximate PI. It was shown that the proposed PIs are less conservative than other fiducial PIs and, for the Poisson case, the HPM-PIs are better than the other PIs in terms of the coverage probabilities. Some examples are presented.