因子服从指数分布的因子分析模型的参数估计研究  

Study on Parameter Estimation of Factor Analysis Model with Exponential Distribution of Factors

作　　者：周国琼 蒋文江[2] ZHOU Guo-qiong;JIANG Wen-jiang(School of Mathematics and Statistics,Zhaotong University,Zhaotong 657000,China;Pan-Asia Business School,Yunnan Normal University,Kunming 650092,China)

机构地区：[1]昭通学院数学与统计学院,云南昭通657000 [2]云南师范大学泛亚商学院,云南昆明650092

出　　处：《云南师范大学学报（自然科学版）》2020年第6期29-34,共6页Journal of Yunnan Normal University:Natural Sciences Edition

基　　金：国家自然科学基金资助项目(11361022).

摘　　要：在假定因子服从指数分布的基础上,构建了一类因子模型,并用基于EM算法的极大似然估计给出模型中的参数估计,其中的E步借助M-H算法来完成,采用的模拟研究方法为MCECM算法,对参数估计的效果进行了模拟分析,结果表明用MCECM算法来确定新因子模型中的参数是合理的.Based on the assumption that the factors obey exponential distribution,a class of factor models is constructed.The maximum likelihood estimation based on EM algorithm is used to estimate the parameters in the model.The E-step is completed by M-H algorithm,and the simulation research method is MCECM algorithm.The simulation analysis of the effect of parameter estimation shows that MCECM algorithm is used to estimate the parameters and the algorithm is reasonable to determine the parameters in the new factor model.

关 键 词：指数分布 EM算法 极大似然估计 M-H算法 

分 类 号：O212.1[理学—概率论与数理统计]