Comparative Analysis of Lasso and Bridge Regression Using Corruption Perception Index and Its Correlates in Nigeria

This study compared LASSO and Bridge regression to determine whether the index of corruption perception is influenced by Human Development, Global Hunger, Global Peace and Consumer Price Indices. The best regression technique in model analysis in handling multicollinearity on Corruption perception and its correlates became imperative. A set of secondary data from 2008 to 2020 from Transparency International, World Bank, Knoema and Country Economy on Corruption Perception, Human Development, Global Hunger, Global Peace and Consumer Price Indexes were used for comparisons based on MSE, R^2, AIC, BIC and VIF. R Software was used to perform regression analysis, while SPSS 22 was used to perform correlational analysis. The test for significance was made at p-value of 0.01 for the standardized variables. The results from the study show that, LASSO regression produced better models with MSE of 0.7748328, R^2 of 0.1608218, AIC value of 8.683596 and BIC of 12.07329 respectively. Though about 16% of variation was explained. Bridge regression produced better uncorrelated results with VIF of 0.9688117 when q = 2, against LASSO with 1.191642. The correlational result from this study shows that, corruption encompasses different relationships and issues on Human development and Global peace are factors influencing corruption in Nigeria.

Keywords:Correlates; Bridge; LASSO; Corruption perception; Index; Regression; Multicollinearity