Bayesian Estimation of Unimodal Distributions

P. Damien, P.W. Laud, A.F.M. Smith
Duke University, Imperial College, Kimberly-Clark Corp., Medical College of Wisconsin

Nov 30 1993

In this paper, we estimate unimodal distribution functions (DFs) from a Bayesian perspective. A DF is assumed to be ogive, that is to say one that switches from convex to concave with an unknown change-point. Monte Carlo methods are described which provide a full Bayesian solution for the modelling of unimodal DFs.


Unimodal distributions, switch-point, convex, concave, gibbs sampling


PDF icon 1994-34.pdf