Examination of Player Positions by Cluster Analysis

Authors : Okan DAĞ, Asım Sinan YÜKSEL, Şerafettin ATMACA

Pages : 43-48

Doi:10.30516/bilgesci.1097014

View : 24 | Download : 10

Publication Date : 2023-03-31

Article Type : Research Paper

Abstract :Today, the football industry stands out among the sports branches. Especially with the development of technology and its integration into football, different tactical understandings and formations emerge. With these developments, the current positions of the players and the other positions they are prone to play can be revealed as a result of the analysis. In this way, club management and technical team aim to establish the best team according to the current budget and tactical game understanding. Therefore, it is very important for the teams to play the players in the right position or to transfer the right player to the required position. In football competitions where 11 players are involved in the game, tactical changes can be made within the game according to the tactical arrangement and tactical understanding of the opposing team, and the player can be played in different positions. In this study, the player data of Turkey and the leagues of Germany, England, France, Spain, Italy, which are considered to be the five big leagues, for the years 2020-2021 were obtained from the website named “whoscored”. In the data set obtained, the players who stayed on the field for a minimum of 1500 minutes were taken as a basis and clustering analysis was performed with the data of 985 players. Players are clustered on four basic positions: goalkeeper, defender, midfielder and attacker. In the study, Expectation Maximization, one of the clustering analysis algorithms, was used and a success rate of 81 percent was achieved.
Keywords : Data Mining, Clustering, Artificial Intelligence