A new hybrid estimator for linear regression model analysis: Computations and simulations
Gladys Amos Shewa, Fidelis Ifeanyi Ugwuowo
Abstract
The Linear regression model explores the relationship between a response variable and one or more independent variables. The parameters in the model are often estimated using the Ordinary Least Square Estimator (OLSE). However, OLSE suffers a breakdown when there is linear dependency among the predictors- a condition called multicollinearity. Several alternative estimators have been suggested as replacements for the OLSE. These include the Kibria-Lukman estimator and the modified ridge-type estimator. In this study, we proposed a hybrid estimator by combining the Kibria-Lukman estimator with the modified ridge-type estimator. The proposed estimator theoretically dominates the existing estimators. The simulation studies and real-life application supports the theoretical findings.