Principal component regression for solving multicollinearity problem

Abstract

Multicollinearity often causes a huge explanatory problem in multiple linear regression analysis. In presence of multicollinearity the ordinary least squares (OLS) estimators are inaccurately estimated. In this paper the multicollinearity was detected by using observing correlation matrix, variance influence factor (VIF), and eigenvalues of the correlation matrix. The simulation multicollinearity data were generated using MINITAB software and make comparison between methods of principal component regression (PCR) and the OLS methods. According to the results of this study, we found that PCR method facilitates to solve the multicollinearity problem.