基于特征样本的变参数模型估计方法研究  

Research on Variable Parameter Model Estimation Method Based on Characteristic Samples

作　　者：李原[1] 王晶 Li Yuan;Wang Jing(School of Statistics,Shanxi University of Finance and Economics,Taiyuan 030006,China;School of Economics and Management,Taiyuan Normal University,Jinzhong Shanxi 030619,China)

机构地区：[1]山西财经大学统计学院,太原030006 [2]太原师范学院经济与管理学院,山西晋中030619

出　　处：《统计与决策》2024年第4期34-37,共4页Statistics & Decision

基　　金：国家社会科学基金资助项目(22ATJ006,16BTJ016);山西省高等学校人文社会科学重点研究基地项目(20210144);山西省“科技战略”专项一般项目(202104031402099)。

摘　　要：文章基于特征样本采样方法,通过对特征样本的重新组合,设计了一种新的变参数模型估计方法,实现经济社会问题中小样本情况下的变参数估计。该方法根据采样顺序将特征样本重组为时间序列各期的样本,通过分期回归实现各期的变参数估计,并对估计结果进行显著性和拟合度检验。该方法简便易行,且符合贝叶斯原理和蒙特卡罗原理,为变参数模型的估计提供了一种新的思路。Based on the characteristic sample sampling method,this paper designs a new variable parameter model estimation method by recombining the characteristic samples to realize the variable parameter estimation in the case of small and medium-sized samples of economic and social problems.The method is used to reorganize the characteristic samples into the ones of each period of the time series according to the sampling order.The variable parameters of each period are estimated by staging regression,and the significance and fit of the estimated results are tested.The method is simple and feasible,and conforms to Bayes principle and Monte Carlo principle,providing a new way to estimate the variable parameter model.

关 键 词：特征样本 变参数模型 计算机模拟 

分 类 号：O212.1[理学—概率论与数理统计]