Studi Metode Regresi Ridge dan Metode Analisis Komponen Utama dalam Menyelesaikan Masalah Multikolinearitas.

Abstract

Multicollinearity is a condition in multiple linear regression where between the independent variables are correlated. Multicollinearity resulted in the regression coefficients from multiple regression analysis to be very weak or can’t provide analytical result that represent the characteristic of the relevant independent variables, and resulted the parameter hypothesis tests using the least square method gives result that are not valid. Indication of multicollinearity problems can be detected with variance inflation factor. Two methods can be used to overcome multicollinearity the method of ridge and principal component analysis. Ridge method aims to reduce multicollinearity by determining the bias estimator but has a smaller variance than the variance of multiple linear regression estimator. The principle component analysis aimed to yield new variables (principle component) which orthogonal to each other and reduce data dimension.