Analiza uporabe složenih statističkih parametara za unaprjeđenje modela strojnog učenja u sportskoj domeni

master's thesis

• Title (english): Analysing the Usage of Complex Statistical Parameters for the Improvement of Machine Learning Models in the Sports Domain
• Author: Dominik Ćurić
• Mentor: Damir Pintar (mentor)
• Committee member: Damir Pintar (predsjednik povjerenstva)
• Committee member: Mihaela Vranić (član povjerenstva)
• Committee member: Luka Humski (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2023-07-13, Croatia

Abstract

U nogometu kao najpopularnijem svjetskom sportu jednostavne statističke katego-
rije poput broja udaraca, broja dodavanja ili broja prekršaja ne pružaju dovoljno dobar
kontekst radnjama koje igrač čini na terenu tijekom utakmice. Zbog toga su potrebne
nove statističke kategorije koje daju bolje i dublje značenje akcija koje igrači čine za
vrijeme utakmice, a zasnovane su na modelima strojnog učenja koji u obzir uzimaju
podatke o događajima (engl. event data). U okviru ovog rada objašnjeni su pristupi
vrednovanja akcija u nogometu te definirani i teorijski objašnjeni Expected Threat i
VAEP modeli. U praktičnom dijelu rada implementirano je više inačica oba modela
te je na temelju rezultata odabrana po jedna inačica svakog od modela i napravljena
usporedba na temelju podataka vezanih uz Englesku nogometnu ligu

Abstract (english)

In football, as the world’s most popular sport, simple statistical categories such as
the number of shots, the number of passes or the number of fouls do not provide a good
enough context for the actions a player performs on the field during a match. For this
reason, new statistical categories are needed that give a better and deeper meaning of
the actions that the players make during the match, and are based on machine learning
models that take into account event data. Within this paper, the approaches to the eva-
luation of actions in football are explained and the Expected Threat and VAEP models
are defined and theoretically explained. In the practical part of the work, multiple ver-
sions of both models were implemented, and based on the results, one version of each
model was selected and a comparison was made based on data related to the English
Football League.

• Keywords: Nogometna analitika; Vrednovanje akcija; Podaci o događajima; Expected Threat model; VAEP model
• Keywords (english): Football analytics; Valuing actions; Event data; Expected Threat model; VAEP model
• Language: croatian
• URN:NBN: urn:nbn:hr:168:321538
• Study programme: Title: Computing; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-10-11 08:41:31