Data Mining in The NBA: An Applied Approach

Abstract

This work aims to analyze the last 6 years of professional basketball data using knowledge discovery tools. The goal is to outline some insight into how teams win and what separates these winning teams from the losing teams. Using Microsoft SQL Server Management Studio, Microsoft Business Intelligence, R and PowerBI, we analyzed regular season game data to find previously unknown trends between winning and losing teams. Using decision trees, naive Bayes and association rules, we will look at defensive and offensive data between teams. The following stats are considered: defensive and offensive rebounds, blocks, steals, turnovers, 2 point shot percentage and 3 point shot percentage. Results show that teams who earn more defensive rebounds win more games while teams that earn higher than average offensive rebounds lose more games. This is because teams that get more offensive rebounds are missing more of their shots.