Bayesian Inference of Generalized Mixed Poisson Linear Models
Based on the Dirichlet distribution,a generalized mixed Poisson linear model is constructed,and the posterior likelihood function is obtained by assuming a prior of unknown parameters.First,obtain the full conditional distribution of unknown parameters by multiplying posterior likelihood with prior.Then,extract unknown parameters using Gbbis algorithm and M-H sampling algorithm to obtain estimated values of parameters,and select regression coefficients.Numerical simulations have verified the effective-ness of Bayesian methods in estimating the parameters of generalized mixed Poisson linear models and the correctness of selecting regression coefficients.
Dirichlet distributiongeneralized mixed Poisson linear modelBayesian estimationvariable selection
NETL
NSTL
万方数据
