Machine learning approaches to predict basketball game outcome

摘要

Sports prediction has always been a spellbinding research area for sports fans, teams, team managers, team players and a growing number of gamblers. Nowadays, companies are spending more effort in machine learning to predict the sport outcomes. Support Vector Machines (SVMs) are powerful techniques that handle classification problems effectively and efficiently. However, SVM models lack in rule generation. So, this examination leads towards the development of Hybrid Fuzzy-SVM model (HFSVM) by integrating fuzzy approach and SVM technique for prediction of the basketball game outcomes that help the coaches, teams and players to enhance their performance. The HFSVM model combines the advantage of both SVM technique and fuzzy approach, which is a unique strength of SVM and rule generation ability of fuzzy approach using fuzzy membership functions. The HFSVM model is compared with SVM model and the empirical results showed that the HFSVM model not only provides better results than SVM model but also provides relatively satisfactory prediction accuracy. Therefore, promising results can be obtained using HFSVM model when analyzing the outcomes of basketball competitions.