ePrints.FRI - University of Ljubljana, Faculty of Computer and Information Science

Predicting performance of football teams using network analysis

Vasil Krstev (2016) Predicting performance of football teams using network analysis. EngD thesis.

Download (1634Kb)


    The goal of this thesis was a detailed analysis of football matches through network analysis and building a predictive model for predicting characteristics of matches based on attributes obtained from such a network. The fields of network analysis and data mining are becoming increasingly popular in the computing world for data knowledge discovery. In the modern world, people are making a lot of measurements on football matches, so we decided to investigate this area in detail, analyse it through network analysis and try to most accurately predict the characteristics of a single game such as the number of points, goals, cards, corners and other. The thesis first presents the methods and techniques for network analysis. Then the algorithms and measures of performance are presented, which are used for the prediction. Next, we present the data, give an example of application of community detection and present the process of building predictive models. What follows is the actual interpretation of predictions and comparison of the effectiveness of predictive accuracy. We have examined the prediction strength for all of the properties obtained from network analysis, but we present only those that gave best results. In conclusion, we briefly compare the results and highlight the main weaknesses. We present possible directions for future work and give guidance on how the predictive model can be further improved.

    Item Type: Thesis (EngD thesis)
    Keywords: football, network analysis, data mining, prediction.
    Number of Pages: 56
    Language of Content: Slovenian
    Mentor / Comentors:
    Name and SurnameIDFunction
    doc. dr. Lovro ŠubeljMentor
    Link to COBISS: http://www.cobiss.si/scripts/cobiss?command=search&base=51012&select=(ID=1537204419)
    Institution: University of Ljubljana
    Department: Faculty of Computer and Information Science
    Item ID: 3529
    Date Deposited: 08 Sep 2016 13:15
    Last Modified: 14 Oct 2016 11:22
    URI: http://eprints.fri.uni-lj.si/id/eprint/3529

    Actions (login required)

    View Item