コンピュータ工学および情報技術ジャーナル

Predicting English Premier League Results using Machine Learning

Yadav A, Sharma A, Gautam A, Bathla G and Jindal R

Football is the most watched, popular sport in the world. The English Premier League is the most popular league in the world. The English clubs playing in the PL have an annual TV rights deal worth over $9 billion, which is the highest paid licensing contract in sports at present. The PL is most followed by billions around the world not only because of its recognition and the big name players playing there, but it’s widely renowned for the sheer uncertainty that it carries. In 2015-2016 season, Leicester City FC came out on top winning against all odds. The betting odds for Leicester winning the PL title were 1/66000 which shows the unpredictability of this league and the results associated with it. In this paper, we have employed Machine Learning techniques to over 6 years of PL dataset by using universal classifiers such as SVM, Logistic Regression, KNN, Decision Trees, etc. The parameters in our dataset have been carefully chosen to help us achieve high accuracy. Moreover, we have used a new approach by employing Poisson’s approach to compare accuracy achieved statistically.