期刊名称：Electronic Journal of Applied Statistical Analysis
电子版ISSN：2070-5948
出版年度：2018
卷号：11
期号：1
页码：340-368
DOI：10.1285/i20705948v11n1p340
语种：English
出版社：University of Salento
摘要：Multicollinearity is a major problem in linear regression analysis and several methods exists in the literature to deal with the same. Ridge regression is one of the most popular methods to overcome the problem followed by Generalized Ridge Regression (GRR) and Directed Ridge Regression (DRR). However, there exist many computational issues in using the above methods. Partial Ridge Regression (PRR) method is a computationally viable approach by selectively adjusting the ridge constants using the cutoff criteria. In this paper, the performance of the Partial Ridge Regression approach has been evaluated through a simulation study based on the mean squared error (MSE) criterion. Comparing with other methods of ridge regression, the study indicates that the Partial ridge regression by cutoff criteria performs better than the existing methods.
其他摘要：Multicollinearity is a major problem in linear regression analysis and several methods exists in the literature to deal with the same. Ridge regression is one of the most popular methods to overcome the problem followed by Generalized Ridge Regression (GRR) and Directed Ridge Regression (DRR). However, there exist many computational issues in using the above methods. Partial Ridge Regression (PRR) method is a computationally viable approach by selectively adjusting the ridge constants using the cutoff criteria. In this paper, the performance of the Partial Ridge Regression approach has been evaluated through a simulation study based on the mean squared error (MSE) criterion. Comparing with other methods of ridge regression, the study indicates that the Partial ridge regression by cutoff criteria performs better than the existing methods.
