GOALALERT: A novel real-time technical team alert approach using machine learning on an IoT-based system in sports

Karakaya A., ULU A., Akleylek S.

Microprocessors and Microsystems, vol.93, 2022 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 93
  • Publication Date: 2022
  • Doi Number: 10.1016/j.micpro.2022.104606
  • Journal Name: Microprocessors and Microsystems
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, Aerospace Database, Applied Science & Technology Source, Communication Abstracts, Computer & Applied Sciences, INSPEC, Metadex, Civil Engineering Abstracts
  • Keywords: Internet of Things, Tactical analysis, Federated learning, Prediction, Machine learning, FOOTBALL
  • Karadeniz Technical University Affiliated: Yes


© 2022 Elsevier B.V.In team sports, the placement of the players before and during the competition/match is very important in terms of tactics. Wrong formation and tactics can directly cause losing the match. In certain parts of the match, the technical team can change the formation of the players according to the tactics. In addition to formation in soccer, there are internal and external parameters to be considered. By using all these parameters, some tactical inferences and predictions can be made during the match. In this paper, a model that provides information and alerts to the technical team about the occurrence of the goal by using machine learning methods on an IoT-based infrastructure is proposed. The data to be obtained from the players through the IoT system is obtained from the FM game because of the difficulty of wearable technologies and the absence of a similar data collection system today. A federated learning concept is used to support the security of data. The proposed model is tested in the data processing area of the IoT system using formation data and some other data, and different variants of discriminant analysis, k-nearest neighbor (KNN), naive bayes, support vector machine (SVM), decision tree and ensemble learning methods. In this concept, AdaBoost, which is one of the ensemble learning methods and optimized for the most suitable parameters, has the highest performance with 87.2%.