有限混合非线性结构方程模型的参数估计  

Paremeters Estimation About the Finite Mixtures of Nonlinear Structural Equation Models

作　　者：梁成侠[1] 傅珏生[1] 

机构地区：[1]苏州大学数学科学学院,江苏苏州215006

出　　处：《苏州市职业大学学报》2008年第1期108-110,114,共4页Journal of Suzhou Vocational University

摘　　要：针对数据不同质现象的广泛存在,将非线性结构方程模型推广到有限混合非线性结构方程模型,用EM算法进行估计,E步引入Gibbs抽样及MH算法,M则采用条件极大的方法。最后,对收敛条件及误差进行估计。Because of the extensive existence of date heterogeneity, we can extend the Nonlinear Structural Equation Models to finite mixtures of Nonlinear Structural Equation Models. EM algorithm is proposed for obtaining the estimates.E-step is completed by a MH algorithm and Gibbs sampler, and the M-step can be completed efficiently by simple conditional maximization. At last, we give the stopping rule and the standard errors.

关 键 词：非线性结构方程模型 有限混合 GIBBS抽样 MH算法 

分 类 号：O212.5[理学—概率论与数理统计]