Title: Addressing complex feature relationships: Harnessing the kernel association coefficient for nonlinear associations Authors:  Kimon Ntotsis - University of Leicester (United Kingdom) [presenting]
Andreas Artemiou - University of Limassol (Cyprus)
Alexandros Karagrigoriou - University of The Aegean (Greece)
Abstract: In the exploration of feature associations, conventional methods often assume linearity. However, real-world data frequently reveals non-linear relationships, necessitating innovative approaches for accurate assessment. The purpose is to introduce a novel kernel association coefficient designed to identify complex associations between features. Comparative analyses against existing coefficients consistently demonstrate higher accuracy. The proposed methodology proves effective across various data distributions and sample sizes, avoiding bias towards linear or non-linear associations. These findings contribute to the advancement of statistical modelling, emphasizing the importance of capturing intricate relationships for inferential, descriptive, or predictive purposes.