Evaluating Machine Learning Varieties for NBA Players' Winning Contribution

Abstract

The NBA brand has been spreading all over the world in recent years. As the league concerns a lot of money and fans, several of researches have been trying to predict its results and winning teams. Through its history, a lot of data and statistics have been collected for every player and going to enrich more details because NBA is developing more than ever. Hence, a big data issue for this kind of predictions deserves to have attention and effort on it. With the availability of such enormous data, it is still complicated to analyze and predict the outcome of match. The study chooses different approaches such as support vector machine regression, polynomial regression and random forest regression for a comparative analysis to discover how individual player's performance influences the team winning rate. The comparison results show that the learning techniques we have adopted exhibit competitive capability in prediction, and give quite consistent performance regardless of complexity in input data features.