Using cooperative networks to analyse behaviour in professional Australian Football

Abstract

ObjectivesReducing the dimensionality of commonly reported complex network characteristics obtained from Australian Football League (AFL) games to facilitate their practical use and interpretability. DesignRetrospective longitudinal design where individual players’ interactions, determined through the distribution and receipt of kicks and handballs, during official AFL games were collected over three seasons. MethodsA principal component analysis was used to reduce the number of characteristics related to the cooperative network analysis. ResultsThe principal component analysis derived two individual-based principal components pertaining to in- and out-degree importance and three team-based principal components related to connectedness and in- and out-degree centralisation. ConclusionsThis study is the first to provide a simplified, novel method for analysing complex network structures in an Australian Football context with both the team- and individual-derived metrics revealing useful information for coaches and practitioners. This may consequently guide opposition analysis, training implementation, player performance ratings and player selection.