Predicting the future performance of soccer players

Gkiastas, Vangelis; Γκιάστας Ευάγγελος

dc.contributor.author	Gkiastas, Vangelis	en
dc.contributor.author	Γκιάστας Ευάγγελος	el
dc.date.accessioned	2022-11-10T09:48:29Z
dc.date.available	2022-11-10T09:48:29Z
dc.identifier.uri	https://dspace.lib.ntua.gr/xmlui/handle/123456789/56115
dc.identifier.uri	http://dx.doi.org/10.26240/heal.ntua.23813
dc.rights	Αναφορά Δημιουργού-Μη Εμπορική Χρήση-Όχι Παράγωγα Έργα 3.0 Ελλάδα	*
dc.rights.uri	http://creativecommons.org/licenses/by-nc-nd/3.0/gr/	*
dc.subject	Sports analytics	en
dc.subject	Machine learning	en
dc.subject	Deep learning	en
dc.subject	Time series	en
dc.subject	Data mining	en
dc.subject	Αθλητικές αναλύσεις	el
dc.subject	Εξόρυξη δεδομένων	el
dc.subject	Μηχανική μάθηση	el
dc.subject	Βαθιά μάθηση	el
dc.subject	Πρόβλεψη χρονοσειρών	el
dc.title	Predicting the future performance of soccer players	en
heal.type	masterThesis
heal.classification	Machine learning	en
heal.classification	Computer Science	en
heal.language	en
heal.access	free
heal.recordProvider	ntua	el
heal.publicationDate	2022-09-02
heal.abstract	Soccer is one of the most widespread and popular sports on the planet and the increasing amount and utilization of data in all aspects of life, could not leave, of course, this field unaffected. The area of sports analytics has attracted a lot of interest in recent years, with applications that affect many aspects of the game. While match outcome predictions, injury preventions, team tactics improvement and betting odds estimations have been widely investigated with different data-driven approaches, future player performance prediction is quite unexplored issue. A perspective of how the athlete's performance will develop in the near or distant future can significantly impact both the athlete and the team on a variety of different levels. This thesis focuses particularly on this issue and is divided in three parts. Firstly, a real-world dataset of elite soccer player games is collected and created, containing individual and team attributes, as well as pre game and other information related to performance. Secondly, it is investigated which variables appear to be more highly associated with the prediction of player's rating in a future game. Finally, the third contribution of this thesis is the forecasting of individual performance in specific future games. Various statistical, machine learning and deep learning models are applied to multivariate time series data with success, producing much better results compared to those of random and naive predictors.	en
heal.advisorName	Karlis, Dimitris	en
heal.committeeMemberName	Ntzoufras, Ioannis	en
heal.committeeMemberName	Fouskakis, Dimitris	en
heal.academicPublisher	Εθνικό Μετσόβιο Πολυτεχνείο. Σχολή Εφαρμοσμένων Μαθηματικών και Φυσικών Επιστημών	el
heal.academicPublisherID	ntua
heal.numberOfPages	90 σ.	en
heal.fullTextAvailability	false