Influence Analysis Of Regression Parameter In Linear Model

This article discusses the Influence Analysis of regression parameter in general linear model and date-delete model. Focus on least squares estimation and ridge estimation, study the Cook distance and Welsch-kuh statistic, analysis the influence ofβin data-delete model, given the relevant conclusions and inferences. Given a new conditional generalized ridge-type estimation by in-depth study of ridge estimate, derived the upper bound of influence in date-delete model by using Welsch-kuh statistic and the nature of eigenvalue.This article introduces the basic concept and research of influence analysis. Then, introduced the matrix, the linear model and the conclusion of parameterβ.Then introduces the basic statistical measure of the influence, Cook distance and Welsch-kuh statistic. Then study the least squares estimation and ridge estimation, set out a new kind of conditional generalized ridge-type estimation, and discuss the relationship between the various estimates. In the and, set out a new kind of conditional generalized ridge-type estimation, derived the upper bound of influence in data-delete model by using Welsch-kuh statistic and give the examples.