Development of English Premier League Team Performance Model

Islami, Muhammad Naufal Alif (2025) Development of English Premier League Team Performance Model. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 05111942000008-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (5MB) | Request a copy

Abstract

Studi ini mengembangkan dan mengevaluasi model pembelajaran mesin yang memprediksi apakah sebuah tim Liga Primer Inggris kemungkinan akan berkinerja baik atau buruk dalam pertandingan mendatang hanya dengan menggunakan informasi yang tersedia sebelum kick-off. Indeks Kinerja Pertandingan pertama kali dibangun pada tingkat pertandingan tim dengan menggabungkan statistik ofensif, defensif, dan kontrol ke dalam satu skor kontinu yang dinormalisasi dalam setiap musim dan dikalibrasi menggunakan bobot berbasis data. Ambang batas tetap pada indeks ini kemudian diterapkan untuk memberi label setiap contoh pertandingan tim sebagai berkinerja baik atau buruk, memberikan target biner yang jelas untuk pembelajaran terawasi. Menggunakan data lima musim Liga Primer Inggris dari 2019/20 hingga 2023/24 yang disusun pada tingkat pertandingan tim, serangkaian fitur pra-pertandingan yang realistis direkayasa. Fitur-fitur ini mencakup performa terkini selama lima pertandingan terakhir, ratarata musim dan musim sebelumnya untuk statistik gol dan tembakan, performa terkini lawan, dan konteks pertandingan dasar seperti status kandang, pekan pertandingan, dan hari sejak pertandingan terakhir. Empat musim pertama digunakan untuk pelatihan dan penyempurnaan, sementara musim 2023/24 dicadangkan sebagai set uji yang belum pernah dilihat sebelumnya
==============================================================================================================================
This study develops and evaluates a machine learning model that predicts whether an English Premier League team is likely to perform or not perform in an upcoming match using only information available before kick-off. A Match Performance Index is first constructed at team match level by combining offensive, defensive and control related statistics into a single continuous score that is normalised within each season and calibrated using data driven weights. A fixed threshold on this index is then applied to label each team match instance as perform or not perform, providing a clear binary target for supervised learning. Using five seasons of English Premier League data from 2019/20 to 2023/24 arranged at teammatch level, a realistic set of pre match features is engineered. These features include rolling recent form over the last five matches, season and previous season averages for goals and shooting statistics, opponent recent form and basic match context such as home status, game week and days since the last match. The first four seasons are used for training and tuning, while the 2023/24 season is reserved as an unseen test set.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Liga Primer Inggris, performa tim, machine learning, rekayasa fitur, klasifikasi tidak seimbang, analitik olahraga ================================================================ English Premier League, team performance, machine learning, feature engineering, imbalanced classification, sports analytics
Subjects:	Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines.
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55201-(S1) Undergraduate Thesis
Depositing User:	Muhammad Naufal Alif Islami
Date Deposited:	02 Feb 2026 03:37
Last Modified:	02 Feb 2026 03:37
URI:	http://repository.its.ac.id/id/eprint/131514

Actions (login required)

View Item