A comparison of some new and old robust ridge regression estimators

Abstract

Ridge regression is used to circumvent the problem of multicollinearity among predictors and many estimators for ridge parameter $k$ are available in the literature. However, if the level of collinearity among predictors is high, the existing estimators also have high mean square errors (MSE). In this paper, we consider some existing and propose new estimators for the estimation of ridge parameter k. Extensive Monte Carlo simulations as well as a real-life example are used to evaluate the performance of proposed estimators based on the MSE criterion. The results show the superiority of our proposed estimators compared to the existing estimators.

Keywords:

• Multicollinearity
• Ridge regression
• Monte Carlo Simulation

2010 AMS Classifications:

• 62J07
• 62G05
• 62F10

Acknowledgments

Authors are thankful to the Editor, Associate Editor, and the anonymous reviewers for their valuable suggestions, which helped to improve the study in its final form.