Indian Journal of Science and Technology
DOI: 10.17485/ijst/2017/v10i13/98678
Year: 2017, Volume: 10, Issue: 13, Pages: 1-8
Original Article
Catalina Garcia*, Roman Salmeron Gomez, Jose Garcia Perez and Maria Del Mar Lopez Martin
Departamento de Metodos Cuantitativos para la Economia y la Empresa, Universidad de Granada, Campus Universitario de La Cartuja, 18071 (Granada), Spain; cbgarcia@ugr.es, [email protected], [email protected], [email protected]
*Author for correspondence
Catalina Garcia
Departamento de Metodos Cuantitativos para la Economia y la Empresa, Universidad de Granada, Campus Universitario de La Cartuja, 18071 (Granada), Spain; [email protected]
Objectives: To analyze the relation of the ridge and raise factors with the squared coefficient of correlation and present, from this relation, a methodology to select the adequate value of the ridge and raise factors. Methods/Statistical Analysis: Two independent variables have been simulated with a given coefficient of correlation between 0.95 and 0.999 and a gap of 0.001 to obtain a dependent variable. Then, it has been selected the value of the ridge and raise factor that leads to values of VIF lesser than 10 for every value of the given coefficient of correlation. Findings: Apart to propose a way to select the raise and ridge factor, it is concluded that the relation between the ridge factor and the squared coefficient of correlation is linear while the relation of the raise factor with the squared coefficient of correlation is potential. Application/Improvements: The procedure presented in this paper allows adequately selecting the value of the ridge and raise factor that mitigates the collinearity. This contribution can be applied in many different fields where collinearity is present.
Keywords: Multicollinearity, Raise Regression, Ridge Regression
Subscribe now for latest articles and news.
