On shrinkage r-estimation in a multiple regression model

From MaRDI portal





The linear model \(X_ i=\beta_ 0+\beta 'c_ i+e_ i\), \(i=1,2,...,n\), is considered where \(\beta_ 0\) and \(\beta '=(\beta_ 1,\beta_ 2,...,\beta_ p)\) are unknown parameters and where the errors are i.i.d. with distribution function F. The parameter \(\beta\) ' is written as \(\beta '=(\beta_ 1',\beta_ 2')\) where \(\beta_ 1'\) and \(\beta_ 2'\) are vectors of \({\mathbb{R}}^{p_ 1}\) and \({\mathbb{R}}^{p_ 2}\) respectively and the authors are interested in the estimation of \(\beta_ 1\) when the parameter \(\beta_ 2\) is close to the value \(\beta^ 0_ 2\). Several estimators are considered: an estimator under the restraint \(\beta_ 2=\beta^ 0_ 2\), an unrestrained estimator, a preliminary test estimator and a James-Stein estimator. These estimators are compared from the point of view of the asymptotic distributional risk.











This page was built for publication: On shrinkage r-estimation in a multiple regression model

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3749975)