Title: An efficient, simulated and modified shrinkage procedure for the simple linear regression model

Authors: Zuhair A. Al-Hemyari; I.H. Hussain

Addresses: Ministry of Higher Education, P.O. Box 82, Postal Code 112, Ruwi, Muscat, Sultanate of Oman ' Department of Mathematics, Baghdad University, Al-Jadria, Baghdad, Iraq

Abstract: In this article, we have proposed two efficient and modified preliminary test shrunken estimators for the response of the simple linear regression model Y, and the regression parameter β when prior guess values of µ_Y and β are available. The expressions for the bias, mean squared error, and relative efficiency of the proposed testimators are derived and studied. Numerical simulations for bias and relative efficiency show that the proposed testimators fare is better than the classical and existing shrinkage testimators. An application of real data is described; comparisons and conclusions are then given.

Keywords: regression line; shrinkage; model modification; numerical simulation; pretest region; bias ratio; relative efficiency; linear regression modelling.

DOI: 10.1504/IJDATS.2015.071369

International Journal of Data Analysis Techniques and Strategies, 2015 Vol.7 No.3, pp.253 - 283

Published online: 22 Aug 2015 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article