基于贝叶斯估计的半参联合模型

Semi-parametric Joint Model Based on Bayesian Estimation

半参数模型可以有效利用样本信息提高精度,同时有效地避免过度拟合.利用B样条方法逼近非参数函数,研究了半参数均值与方差联合模型的贝叶斯估计问题.通过吉布斯抽样与MH算法来得到模型中未知参数的贝叶斯估计.最后经过模拟说明了所提出的贝叶斯分析方法是有效的.

Semi-parametric model can effectively improve the accuracy of sample information and avoid over-fitting.The spline method is used to approximate the nonparametric function in this paper,and the Bayesian estimation of the semi-parametric model is studied.Then,Bayesian estimation of unknown parameters in the model is obtained by Gibbs sampling and MH algorithm.Finally,the simulation results show that the proposed Bayesian method is effective.