Variable Selection with Metaheuristic Methods

• Activity Number: 173 - Recent Advances in Statistical Learning and Missing Data Handling
• Type: Contributed
• Date/Time: Tuesday, August 4, 2020 : 10:00 AM to 2:00 PM
• Sponsor: Korean International Statistical Society
• Author(s): Myung Soon Song* and Francis J Vasco and Yun Lu and Kyle Callaghan
• Companies: Kutztown University of Pennsylvania and Kutztown University of Pennsylvania and Kutztown University of Pennsylvania and Kutztown University of Pennsylvania
• Keywords: variable selection; metaheuristics; optimization; Jaya

Abstract

Variable selection (or feature selection) has been one of old topics in regression models. Beside many classical approaches, some metaheuristic approaches from the optimization research such as GA (Genetic Algorithm) or SA (Simulated Annealing) have been developed so far. These methods have a considerable advantage to deal with high dimensional problems over the conventional methods, but they must control associated fine-tuning parameters, which is very hard in practice. In this article, JAYA, one of the parameter-free approaches will be suggested and explored. Many methods such as GA, TBO, and JAYA will be compared to one another with the results from real-world datasets.