Model Selection and Post-estimation via Pretesting: Ridge Regression

Abstract

The goal of this study is to improve the efficiency of parameter estimation in a gamma regression model, when there was uncertainty about the quality of subspace information and multicollinearity was present. Ridge-type of pretest estimation strategy was applied and a Monte Carlo simulation was conducted to evaluate the proposed estimators. These estimators outperformed the classical ridge regression estimator. The suggested strategy was applied to a real dataset to test the practicality of the estimators.