Parameter estimation and heterogeneity research based on mixed Gamma-Poisson distribution model
There are few parameter estimation methods for continuous-discrete mixed distribution,and most of them have the problem of inverse of high-dimensional matrix or low estimation efficiency.In this paper,the minorization-maximization(MM)algorithm and assembly decomposition technique are applied to parameter estimation of Gamma-Poisson mixed distribution.The purpose is to separate and assemble the high-dimensional objective function into a series of linear combinations of low-dimensional functions.Then the difficulty of finding the inverse of high dimensional matrix can be avoided effectively.A series of simulation studies show that MM algorithm and its assembly decomposition have strong accuracy and stability in parameter estimation of Gamma-Poisson mixed distribution model.Applying the Gamma-Poisson mixed distribution model to the data of divorce duration in Belgium,it is found that the Gamma-Poisson mixed distribution model has a good fitting effect on this set of heterogeneous continuous-discrete data.
Gamma-Poisson mixed distributionminorization-maximization algorithmheterogeneityBelgium divorce statisticsassembly and disassembly technique
万方数据
维普
