Performance of Beta Ridge Regression Estimator in Addressing Multicollinearity within Beta Distribution

Abstract

Beta Ridge Regression (BRR) is a ridge method applied in the beta regression model used to overcome the problem of multicollinearity, which is a condition in which the independent variables in the regression model have a high correlation.  This problem can cause parameter estimates to be unstable and less accurate.  This study aims to determine the performance of BRR estimator in overcoming multicollinearity in simulated data with small sample size. The analysis is done by comparing the estimation results based on the Mean Squared Error (MSE) and Mean Absolute Error (MAE) values.  The results show that the proposed BRR estimator has superior performance compared to the Maximum Likelihood Estimation (MLE) method, by producing lower MSE and MAE values than MLE.

Keywords: Multicollinearity, Beta Ridge Regression, Beta Distribution, Simulated Data, Mean Squared Error, Mean Absolute Error