Graph-based approaches for analyzing team interaction on the example of soccer

We present a graph-based approach to analyzing player interaction in team sports. A simple pass-based representation is presented that is subsequently used together with the PageRank algorithm to identify the importance of the players. Aggregating player scores to team values allows for turning our approach into a predictor of the winning team. We report on empirical results on five German Bundesliga seasons.
© Copyright 2015 Machine Learning and Data Mining for Sports Analytics ECML/PKDD 2015 workshop. Published by Department of Computer Science, KU Leuven. All rights reserved.

Bibliographic Details
Subjects:
Notations:technical and natural sciences sport games
Tagging:data mining
Published in:Machine Learning and Data Mining for Sports Analytics ECML/PKDD 2015 workshop
Language:English
Published: Leuven Department of Computer Science, KU Leuven 2015
Online Access:https://dtai.cs.kuleuven.be/events/MLSA15/papers/mlsa15_submission_3.pdf
Pages:15-22
Document types:article
Level:advanced