B1486
Title: Predicting handball games with machine learning and teams strengths statistics
Authors: Florian Felice - University of Luxembourg (Luxembourg) [presenting]
Christophe Ley - University of Luxembourg (Luxembourg)
Abstract: A statistically enhanced learning (aka. SEL) model is presented to predict handball games. The machine learning model augmented with SEL features outperforms state-of-the-art models with an accuracy beyond 80\%. It is shown how the strengths of teams are estimated, representing the SEL covariates. Different models are compared and evaluated on female clubs' data to assess their predictive capabilities. It is shown that SEL variables appear as the most important features of the model. Finally, it is shown that explainability methods can help identify important drivers of the scored goals by a team. This can be used as a new predictive tool for team coaches to adjust their strategies in view of upcoming matches.
