Klassifizierung von Interaktionsprozessen im Sportspiel mit Hilfe selbstorganisierender Karten

Research output: Book/ReportDissertationsResearch


  • Andreas Grunz

Research units


In the present work a method for classification of team tactical interaction processes based on Self-organizing maps is developed. The classification learns the different patterns from examples that an expert has categorized in advance. The developed procedure can support the evaluation with performance indicators or serve as the basis for a new indicator. First the current state of research is presented from the point of view of sports science, informatics and sport informatics. Afterwards the Self-organizing map (SOM) developed by Kohonen (1982), the Dynamically Controlled Network (DyCoN) developed by Perl (2001b) and the Merge Self-organizing map (MSOM ) developed by Strickert und Hammer (2003) are introduced. As first subgoal the complex team tactical interaction process in sport games is modeled with Self-organizing maps. The dynamics of the interac- tion process must be preserved by the model. Based on the results of the previous work a hierarchy of DyCoNs is developed in order to classify tactical interaction patterns in soccer. An evaluation of the new approach is carried out on the position data of the 2006 World Cup Finals. Finally the developed method is applied to a basketball game. In addition the MSOM is adapted as an alternative method for classification. The results of the two methods are finally compared.
Keywords: analysis, movement patterns, soccer, self-organizing map, ga- me analysis, tactics, team
Original languageGerman
Place of PublicationKöln
PublisherDeutsche Sporthochschule Köln
Number of pages53
Publication statusPublished - 2021

ID: 5600552


View graph of relations