基于经验似然的AR(p)模型的统计诊断  

Diagnostic for Auto-regression Time Series Models based on Empirical Likelihood Method

作　　者：张敏珏[1,2] 冯予[1] 

机构地区：[1]南京理工大学理学院,江苏南京210094 [2]池州学院数学与计算机科学系,安徽池州247000

出　　处：《数理统计与管理》2014年第2期286-295,共10页Journal of Applied Statistics and Management

基　　金：国家社会科学基金项目(09BTJ004)

摘　　要：本文基于经验似然方法对AR(p)模型进行统计诊断,文章首先给出p阶自回归模型的广义估计函数并对模型参数进行估计,然后运用数据删失、局部影响分析和伪残差方法对AR(p)模型进行统计诊断,最后通过实证来说明该诊断方法的有效性.In this paper, we study diagnosis for auto-regression time series model based on empirical likelihood method. First, the generalized estimating function is given and estimators of the parameters are obtained. Then, based on case deletion, local influence analysis and pseudo-residuals method, the problem of outliers in auto-regression time series models is selected. Last, an example is given to illustrate the validity of diagnosis.

关 键 词：经验似然 广义估计函数 局部影响分析 统计诊断 

分 类 号：O212[理学—概率论与数理统计]