Gutenberg Open


Titelanzeige

Personen: Winter, Christian (Autor) 
Rasche, Christian (Autor) 
Pfeiffer, Mark (Autor) 
  
Titel: Linear vs. non-linear classification of winners, drawers and losers at FIFA World Cup 2014
  
Quelle: Science and medicine in football. Bd. 1. H. 2. London : Taylor & Francis Group. S. 164 - 170
Erscheinungsjahr:    2017
ISBN / ISSN: 2473-4446 ; 2473-3938
URL der Originalveröffentlichung doi:10.1080/24733938.2017.1283435
Bemerkung: Nachweis im WorldCat
  
Dokumentart:
Zeitschriftenaufsatz Zeitschriftenaufsatz
Sprache: Englisch
Open Access:
Personen der Universität:    Winter, Christian  In UnivIS suchen  ; Rasche, Christian  In UnivIS suchen ; Pfeiffer, Mark  In UnivIS suchen 
Einrichtung: Institut für Sportwissenschaft
DDC-Sachgruppe:    Sport
ID: 56327  Universitätsbibliothek Mainz
Hinweis:
Informationen zu den Nutzungsrechten unserer Inhalte Informationen zu den Nutzungsrechten unserer Inhalte
Abstract: Purpose: Tactical analyses to distinguish between football teams that were more or less successful have been conducted up to now only by means of linear methods (like discriminant analysis). Concerning the non-linear relationships between performance related conditions, performance and success in sports games, a non-linear method could be more appropriate.

Methods & Results: Therefore, all knockout matches played during FIFA World Cup 2014 were analysed using tactical metrics. Results lead to 4 different dimensions (Transition play, Creating scoring opportunities, Defense and Scoring) from which especially the latter was essential to differentiate between winners, drawers and losers. Linear discriminant analysis identified 43.30% of the cases correctly whereas a non-linear artificial neural network (ANN) lead to a successful classification of 57.85% all together.

Conclusion: Considering that the differences concerning the tactical behaviour between more and less successful teams were small due to the homogeneous level of performance, the results of the discrimination by means of artificial neural networks indicate non-linear to be more adequate compared to linear methods for future analyses of sports games.

Practical Implications: For sport scientists: The results of our study indicate a superior classification by means of the non-linear model of Artificial Neural Networks compared to the linear model of Discriminant Analysis. Therefore, it is suggested that these kind of analyses may be suited better to model the complex relations between performance indicators and match success in sports games.

For sport practitioners: It is shown that the behaviour during scoring is by far most important for differentiating between more or less successful teams in FIFA World Cup 2014. Different teams used different playing styles, that were equally successful, to create scoring opportunities. But there is no way around a good ratio between goals and scoring opportunities. This implies that either successful teams take shots only from situations that are more promising or they are more successful in scoring in equally promising opportunities.
   
  
Verfügbarkeit prüfen:    Rechercheportal Mainz: 2473-4446/2473-3938
  Elektronische Zeitschriftenbibliothek (EZB): 2473-4446 oder 2473-3938
 


Im   
Impressum