A New General Ridge-Type Estimator in Linear Regression Models

Accepted - January 2024

Abstract

In linear regression models, researchers develop new biased estimators to alleviate the effects of multicollinearity, rather than using the Ordinary Least Squares (OLS) method which is affected by multicollinearity. However, we observed that some of these proposed estimators can be included in a general class. In this article, we define a general class of these estimators called Ridge-type Estimators (RTE). The crucial feature of RTE is that it has the advantageous properties of both Ridge and Liu Estimators and also contains a functional relationship between the biasing parameters. Alternative approaches are presented to determine the optimal functional relationship. The superiority of RTE over other biased estimators is investigated under the matrix mean square error criterion. Two separate Monte Carlo simulation studies are conducted to compare the performance of the considered biased estimators. A numerical example is given to demonstrate the superiority of the proposed estimator over other biased estimators.