Football match outcome prediction by applying three machine learning algorithms

Abstract

Football prediction has become an interesting problem, and researchers are trying to find solutions that are different from each other. The idea of modeling and building a smart model that studies the available data and expects the outcome of the game has become more popular and widespread in the past few years. Professionally different predictive methods have been developed to assess the characteristics that cause a soccer team to lose a match or win a match. Machine Learning (ML) is one of the branches of artificial intelligence that is concerned with designing algorithms that allow computers to have the advantage of learning without programming the rules for each issue. These algorithms consist of a series of commands, and instructions necessary to direct the machine or computer to how the tasks should be carried out, as the algorithms play the role of the mastermind in the machine because of its polarization of data, collection, analysis and finally relying on the analyzed data to determine how the task should be performed. The algorithms used in machine learning rely on a set of graphical models and decision tools such as the decision tree, natural language processing, and artificial neural networks for the task of automating analyzed data and processing; thus, motivating the machine to make decisions and carry out the tasks assigned to it with precision and ease. This paper implements three ML algorithms; Decision Forest (DF), Artificial Neural Network (ANN) and Support Vector Machine (SVM) to predict the match result. The result showed that DF achieved the highest accuracy of 89%, ANN achieved the accuracy of 72%, and SVM achieved the accuracy of 70%. It is worth noting that the prediction results are obtained based on 10-folds cross-validation.