Prediction of Turkish Super League Match Results Using Supervised Machine Learning Techniques

Abstract

Over the last number of years, sports analytics has become more popular in supporting personnel decisions, evaluating player and team performances, and predicting game results in various sports. One of the most traditional sports, football is also modernizing its ways based on sports analytics techniques. The purpose of this study is to propose a football match prediction model for Turkish Super League (TSL) using supervised machine learning techniques. To do this, based on the TSL data of last five years (2013 to 2018), game result prediction models were established using classification techniques including logistics regression, linear and quadratic discriminant analyses, K-nearest neighbors, support vector machines, and random forests. An ensemble of 10 models based on seven different techniques is suggested.