Olugbenga Ojo & Nurudeen Alabi

This research presents a comparative analysis of the two most widely used shrinkage methods Ridge and Lasso Regression with Leave-One-Out Cross Validation (LOOCV) used in the determination of tuning parameters. The adoption of Mean Square Error (MSE) as a statistical loss function, where the exploration aims to single-out the optimal regression model in the face of multi-collinearity challenges was ensured. These regression model estimation techniques were compared using data on selected macroeconomic variables such as Gross Domestic Product, Transportation and Storage, Information and Communication, Construction, Trade, and Oil Refining in Nigeria. Empirically, the RIDGE and RIDGE_LOOCV techniques exhibited a superior efficiency with MSE values (0.0174 and 0.0054, respectively) compared to LASSO and LASSO_LOOCV with MSE values (0.1081 and 0.0436, respectively) afterwards, showing their enhanced performance states. Thus, the RIDGE and RIDGE_LOOCV regression techniques effectively optimized the model estimation process within the study framework. While in overall, they deduced as techniques that can best address the multi-collinearity issues within the applicability of macroeconomics. KEYWORDS: Ridge Regression, Lasso Regression, LOOCV, Mean Square Error, Macroeconomic0150