Understanding Complexity: A Machine Learning Approach to Exploring Playing Styles in Australian Football

Abstract

The thesis introduced a novel clustering framework utilising a composite cluster assessment index informed by subject matter expert knowledge to guide the selection of optimal cluster hyper-parameters. The framework defined five offensive, five defensive, and six transitional team playing styles and six offensive, six defensive, and four transitional player playing styles implemented in the 2021 – 2023 AFL seasons. Further analysis of team playing styles and individual playing performance uncovered deeper insights to assist coaching decisions.