An analytical approach to Bayesian generalized linear models

点击翻译按钮获取中文摘要

The following three major topics were investigated.(1) Noting the equality satisfied by the optimal predictive density, we discussed a Bayesian model selection criterion.(2) We derived duality in the Bayesian estimation from the analytical property of having a conjugate prior.(3) We investigated a linear mixed model that is based on a location family having a conjugate prior.As an application study we analyzed the rainfall data of East Australia using a generalized linear model whose underlying distribution is the Tweedie distribution.