Title: Penalized Quantile Regression for Enhancing the Accuracy of Variable Selection
Abstract: The penalized regularization regression methods have been extensively studied in the literature, but it is still difficult to generalize them, particularly to applications characterized by heterogeneity and collinearity. Due to that, hybrid statistical and machine learning methods have been developed. In this context, two methods are elastic-net and adaptive lasso regression, where the weights are based on a quantile regression estimator. We have applied those methods using several scenario of simulation and application of crude oil prices in European Union. Moreover, the methods performance is verified by the criteria RSS, RMSE, MAE, MAPE, and MASE. Overall, the finding of both simulation and real data application show that the quantile elastic-net regression generally outperforms all other methods with respect to all measures.