Πανεπιστήμιο Μακεδονίας
The applications of rating methods combined with machine learning techniques
Abstract
dc:descriptionThis dissertation examines the rating systems and their applications in various fields. Most of the systems studied were mainly developed for the sports field. Due to this fact, the main application developed in this thesis is related to the sports field and focuses on the predictions of the outcomes of English Premier League games by utilizing rating systems and machine learning techniques. The resulting prediction models from this application are derived either through team rankings, statistical methods, or a combination of machine learning techniques. Our research findings from the integration of rating systems with machine learning techniques are highly encouraging in terms of predictive quality and risk-adjusted investment opportunities. Moreover, three distinct applications have been developed in other fields than sports, where the rating systems are utilized with or without combining machine learning approaches. The first application concerns the ranking of domain names, the second deals with financial management and optimization contexts, and the third focuses on user preference ratings and recommendations. This thesis introduces two rating systems. The first is a novel rating system that deals with the rating and ranking of soccer teams by taking into account the outcome of games, the margin of victory, and the shooting accuracy of each team. The second system is more generalized and applicable in various fields for rating/ranking where the basic idea behind the method is the WSM (Weighted Sum Method) and in fact, it is a modified version of the MAUT (Multi-Attribute Utility Theory) / MAVT (Multi-Attribute Value Theory). The effectiveness of our proposed systems is evaluated in the main application of this thesis where they have been compared to other established systems with satisfactory results. Additionally, the second system is utilized in the second distinct application that focuses on optimization contexts in finance. Finally, an open-source software dedicated to the implementation of rating/ranking systems with applications and examples in sports and other fields was developed and is provided.
Degree
thesis:*- Grantor dc:publisher
- Πανεπιστήμιο Μακεδονίας
- Year dc:date
- 2023
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Talattinis, Kyriacos
- Ταλαττίνης, Κυριάκος
Subjects
dc:subject × 22- Μέθοδοι βαθμολόγησης
- Συστήματα κατάταξης
- Μηχανική μάθηση
- Μοντέλα πρόβλεψης
- Εφαρμογές βαθμολόγησης
- Rating methods
- Ranking systems
- Machine learning
- Prediction models
- Rating applications
- Επιστήμη Ηλεκτρονικών Υπολογιστών και Πληροφορική
- Φυσικές Επιστήμες
- Τεχνητή νοημοσύνη
- Επιστήμη ηλεκτρονικών υπολογιστών και Πληροφορική, άλλοι τομείς
- Μαθηματικά
- Εφαρμοσμένα μαθηματικά
- Computer and Information Sciences
- Natural Sciences
- Artificial Intelligence
- Computer and Information sciences, miscellaneous
- Mathematics
- Applied Mathematics
Rights
- Language dc:language
- eng
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/55275