Premier League Match Result Prediction using Machine Learning

Abstract

The ease of the accessibility of Internet and the popularity of Machine Learning have been the prime reasons in the increase of Sports Analysis and Betting. Football being the most popular sports in the that is played in over 200 countries world is regarded as much more dynamic and complex when compared to other prevailing sports and this makes football an interesting field for research. For the development of prediction systems several methodologies and approaches are being used. In this project we predict the result of a Premier League match given a home team and an away team. The predictions are made based on various important attributes that consists of data from previous seasons of Premier League. These important attributes are very likely to decide the outcome of a match. For the purpose of predictions, we use three different algorithms namely, Logistic Regression, XGBoost and Support Vector Machines and then we select the best algorithm out of these three to predict that appropriate label. The application of these models are done on real team data and results of fixtures are gathered from http://www.football data.co.uk/ for the seasons ranging from 2003/04 to 2018/19