SOME NEW ADJUSTED RIDGE ESTIMATORS OF LINEAR REGRESSION MODEL

The ridge estimator for handling multicollinearity problem in linear regression model requires the use the biasing parameter. In this paper, some new adjusted ridge parameters which do not require the biasing parameter are proposed. The performances of the proposed Adjusted Ridge Estimators are compared with a recently proposed Adjusted Ridge Estimator, Generalized Ridge Regression Estimator (GRRE), Ordinary Ridge Regression Estimator (ORRE) and Ordinary Least Square estimator (OLSE) via Monte Carlo study by counting the number of times each estimator has smallest Mean Square Error (MSE) in ten thousand (10,000) replications. The proposed Adjusted Ridge Estimator is most efficient especially when multicollinearity is severe and the error variance is high