The release of NBA player tracking data greatly enhances the granularity anddimensionality of basketball statistics used to evaluate and compare playerperformance. However, the high dimensionality of this new data source can betroublesome as it demands more computational resources and reduces the abilityto easily interpret findings. Therefore, we must find a way to reduce thedimensionality of the data while retaining the ability to differentiate andcompare player performance. In this paper, Principal Component Analysis (PCA) is used to identify fourprincipal components that account for 68% of the variation in player trackingdata from the 2013-2014 regular season and intuitive interpretations of thesenew dimensions are developed by examining the statistics that influence themthe most. In this new high variance, low dimensional space, you can easilycompare statistical profiles across any or all of the principal componentdimensions to evaluate characteristics that make certain players and teamssimilar or unique. A simple measure of similarity between two player or teamstatistical profiles based on the four principal component scores is alsoconstructed. The Statistical Diversity Index (SDI) allows for quick andintuitive comparisons using the entirety of the player tracking data. As newstatistics emerge, this framework is scalable as it can incorporate existingand new data sources by reconstructing the principal component dimensions andSDI for improved comparisons. Using principal component scores and SDI, severaluse cases are presented for improved personnel management.
展开▼
