Using Machine Learning to Analyze Football Teams and Predict the Outcome of a Football Match

Adish Golechha; Akshat Muke

doi:10.37591/joces.v12i1.905

Abstract

Football, as one of the most popular sports on the planet, has always attracted a large number of fans. Over 150 million men and women of all ages play it in over 200 countries. Modern football has seen a paradigm shift from being just one of the most physical sports to now being one of the most complex sports due to the involvement of multiple new factors such as home games, away games, form of an individual, total shots taken, total shots blocked, weather conditions etc. In recent years, the importance of these statistics in football and its usage has been seen by all. In this study, football statistics have been used to predict the outcome of football matches via various machine learning algorithms as well as by analyzing them. A thorough examination of these statistics enables a manager to devise strategies for team management as well as upcoming matches. To calculate accuracy, three Machine Learning algorithms were used: Logistic Regression, Support Vector Machine, and Multinomial Nave Bayes. All three of them were compared and it was observed that the most efficient accuracy was that of SVM which was (61.29%). It was also noted that the accuracy achieved by Logistic Regression (60.9%) was very close to that of SVM.

Abstract

Keywords