Talks

Equivalent Poisson and iid sampling models: A Bayesian view.

111
reads

Albert Lo 
2010-03-26 
13:00 - 14:00 
402 , Freshman Classroom Building



Equivalent Poisson and iid sampling models: A Bayesian view. Abstract. The estimation of the intensity measure of a non-homogeneous Poisson process and the estimation of a distribution from an iid sample are equivalent problems. This raises the question if this view continues to prevail in the Bayesian approach. It is shown that the fundamental result that the Dirichlet processes is a conjugate family of priors for iid sampling from an unknown distribution is equivalent to the result that the Gamma processes form a conjugate family of priors when sampling from a non-homogeneous Poisson process. Connections of this equivalency, as well as the usual prior-posterior updating in Bayesian analysis, to the theory of Palm distributions in random measure theory are also discussed.