Metode Penanganan Multikolinieritas pada RLB: Perbandingan Partial Least Square dengan Ridge Regression

Abstract

Multicolinierity between variable predictor in multiple regression is assuming violation  for ordinary least square estimator (OLS). Ridge Regression (RR) and Partial Least Square Regression (PLSR were used to handle the multicolinierity problem. RR modify OLS by adding subjective bias consatant, while PLSR, generalize and combine Principal Component Analisis and multiple regression. The efficiency of these two methods will be compared based on the value of RMSE. This study simulated generating data in different level of multicolinearity, the number of variabel, and number of observation were controlled. This study results that, overall, both method equally efficient.