Napovedovanje izbire akcije na četrtem poskusu v ameriškem nogometu

Blazina, Iza (author), Vračar, Petar (mentor)

V diplomski nalogi smo analizirali problem napovedovanja odločitev trenerjev v situacijah četrtega poskusa (4th down) v ameriškem nogometu z uporabo metod strojnega učenja in podatkov play-by-play. Iz tekem lige NFL v sezonah 2018–2024 smo pripravili podatkovno zbirko situacij četrtega poskusa ter oblikovali atributni prostor, ki je vključeval značilnosti trenutnega stanja igre, vremenske pogoje, stavniške napovedi in statistike ekip iz preteklih tekem. Za napovedovanje treh možnih odločitev trenerjev smo uporabili večrazredne klasifikacijske modele, pri čemer je najboljšo uspešnost dosegel model XGBoost s klasifikacijsko točnostjo 89,8 % in mero F1 0,857. Analiza pomembnosti atributov je pokazala, da so za napovedi modela najpomembnejši položaj na igrišču, pričakovane točke, razdalja do prvega poskusa in preostali čas igre. Rezultati so pokazali skladnost med napovedmi modela in uveljavljenimi strateškimi vzorci odločanja v ameriškem nogometu.

• Language: Slovenian
• Keywords: strojno učenje, večrazredna klasifikacija, analiza športnih podatkov, četrti poskus, NFL
• Work type: Bachelor thesis/paper
• Organization: FRI - Faculty of Computer and Information Science
• Year: 2026
• PID: 20.500.12556/RUL-187601
• Publication date in RUL: 11.09.2026

Secondary language

• Language: English
• Title: Predicting the Choice of Play on Fourth Down in American Football

Abstract

In this thesis, we analyzed the problem of predicting coaches’ decisions in fourth-down situations (4th down) in American football using machine learning methods and play-by-play data. Using NFL games from the 2018–2024 seasons, we constructed a dataset of fourth-down situations and designed a feature space that included characteristics of the current game state, weather conditions, betting predictions, and team statistics from previous games. To predict the three possible coaching decisions, we applied multiclass classification models, with XGBoost achieving the best performance, reaching a classification accuracy of 89.8 % and an F1 score of 0.857. Feature importance analysis showed that the model relied primarily on field position, expected points, distance to the first down, and remaining game time. The results demonstrated consistency between the model predictions and commonly observed strategic decision-making patterns in American football.

• Keywords: machine learning, multiclass classification, sports analytics, fourth down, NFL