随机约束线性回归模型参数估计的影响分析

The Influence Analysis in Stochastic Constrained Linear Regression Model

讨论了带随机约束的线性回归模型的影响分析及强影响点的探测问题 .利用数据删除法 (casedele tion) ,导出了在单点剔除模型下参数的混合估计与原模型下对应估计的相互关系 ,并利用Cook统计量、Welsch-kuh统计量和协方差比统计量 ,分别给出了剔除单个数据点对参数估计的影响度量 .通过数据模拟例子 ,说明了此方法的有效性 .

The influence analysis and the detections of the influential problems in the Linear Regression Model with linear stochastic restrictions are discussed.Case deletion method is used to derive the relations of the mixed estimators and the original estimators in single case deleted model.Influence measures Cook statistic,Welsch kuh statistic and covariance ratio statistic of parameter estimators are obtained.In addition,a simulate data example is provided for illustration purpose,which show this method is very useful in practice.