Football Prediction System using Gaussian Naïve Bayes Algorithm

Abstract

Soccer (or, more colloquially, football) is among the most popular sports around the globe, with a thriving economy valued at more than $400 billion and billions of supporters (estimated) worldwide. Predicting match results has always piqued people's interest, and studying game results has shown to be extremely beneficial for corporate success and player development. There are many machine learning approaches for game prediction, however, it is believed the Bayesian approach could be very helpful in this scenario (given reliable historical data). The proposed model has been enforced on authentic squad information including match results collected from kaggle.com and other websites like Sofifa.com. Observations indicate that the Gaussian Naive Bayes Approach is capable of predicting the results of a football match with an accuracy of 85.43%, which is a bit higher than the 79.81% accuracy that is achieved using the Decision Tree Classifier