Detecting drivers of basketball successful games: an exploratory study with machine learning algorithms

This paper aims to identify the drivers leading to victory for basketball matches in NBA, the american National Basketball Association. Firstly, a dataset containing box scores and Dean's four factors for regular seasons from 2004-2005 to 2017-2018 has been prepared. Then, box scores and four factors have been used as classification independent variables, to predict the winner of matches involving the Golden State Warriors team. Both CART and Random Forests machine learning techniques have been applied, and quality of fitting analyzed. Variable importance of fitted models has been studied to identify success drivers showing how, for Golden State Warriors, defense is a key factor to win a game. At last, these models are shown to be suitable for coaching staff in game preparation, and CART models are shown to be valuable on the basketball court for match interpretation.