Implementation and validation of new optimization methods by genetic algorithm for two‐parameter ridge estimator

• Published in: Concurrency and computation Vol. 33; no. 9
• Main Authors: Tekeli, Erkut, Özbay, Nimet, Kaçiranlar, Selahattin
• Format: Journal Article
• Language: English
• Published: Hoboken, USA John Wiley & Sons, Inc 10-05-2021; Wiley Subscription Services, Inc
• Subjects: Estimators; genetic algorithm; Genetic algorithms; multicollinearity; Optimization; Optimization techniques; Parameter estimation; Regression analysis; Regression models; ridge estimator; two‐parameter ridge estimator
• ISSN: 1532-0626; 1532-0634
• DOI: 10.1002/cpe.6088

Two‐parameter estimators have increasing usage in the linear regression model concerning mitigating the problem of multicollinearity. In this type of biased estimators, two different parameters contribute to the solution of two different problems. Previously defined two‐parameter ridge estimator (TPRE) assures considerable merits in this context. This estimator eliminates unfavorable effects of multicollinearity as well as improves the coefficient of multiple determination for the linear regression model. Concerning the TPRE, both the mean square error comparisons and some conventional selection methods for the biasing parameters are available in the literature.
In this article, we mainly focus on the simultaneous estimation of the biasing parameters of the TPRE through some new optimization techniques by genetic algorithm. To observe validation of the new approaches, we perform a numerical example in addition to a Monte Carlo study. The outcomes of these runnings prove the dominance of the new approaches in comparison to existing techniques in the literature.