A0473
Title: Marginalized LASSO in the difference-based partially linear model for variable selection
Authors: Mina Norouzirad - Center for Mathematics and Applications (NovaMath) (Portugal) [presenting]
Ricardo Moura - Center for Mathematics and Applications (Portugal)
Mohammad Arashi - Ferdowsi University of Mashhad (Iran)
Filipe Marques - University of Lisbon (Portugal)
Abstract: The difference-based partially linear model is suitable for regression when both linear and nonlinear predictors are present in the data. Optimizing the weights using the difference-based method presents challenges in variable selection, particularly with low-variance predictors. A novel methodology based on marginal theory is proposed to address these mixed relationships effectively, emphasizing variable selection through a marginalized LASSO estimator with a penalty term that is less severe and related to the difference order. Comprehensive simulation experiments evaluate the performance of the proposed technique in estimation and prediction compared to the LASSO estimator. Additionally, the bootstrapped method is employed to assess the performance of the proposed prediction method using the King House dataset.
