Online Program Home
My Program

Abstract Details

Activity Number: 177 - Big Data and Computationally Intensive Methods
Type: Contributed
Date/Time: Monday, July 29, 2019 : 10:30 AM to 12:20 PM
Sponsor: Section on Statistical Computing
Abstract #305153
Title: Hybrid Ridge-Lasso Regression
Author(s): Saeed Aldahmani* and Taoufik Zoubeidi
Companies: UAE University and UAE University
Keywords: Linear regression; Lars; Lasso; Ridge regression

This paper introduces a new less biased regularization method called HRLR for dealing with the problem of linear regression when the number of variables (p) is greater than the observations (n) in the data set, i.e., n< p. The proposed method is able to give less biased estimates for important parameters unlike existing methods such as ridge, Lasso and Lars. The method iteratively selects significant variables (important ones) by using a pre-selection method and awarding a zero penalty to these variables. This selection process guarantees less biased coefficient estimates for the set of important variables and helps in findings the significant predictors that relate to the response. This is likely to improve the accuracy of variable selection and prediction. Both numerical and analytic discussions are used to verify the effectiveness of the proposed method. The results of the simulation and data analysis are contrasted with those of ridge, Lars and Lasso. The findings show that HRLR outperforms the other methods when n< p.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2019 program