Application of Principal Component Analysis of Distribution in Sport Analytic

• Activity Number: 537
• Type: Contributed
• Date/Time: Wednesday, August 12, 2015 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Marketing
• Author(s): Sun Makosso-Kallyth* and Brahim Brahim
• Companies: Degroote-Pain Centre-McMaster University and BigData Visualizations
• Keywords: Data visualization ; Big data ; Sports analytics ; Principal component analysis ; Distributional data

Abstract

The use of exploratory data analysis in sport has increased recently. This new interest is motivated by economic reasons, marketing issues, the necessity of coaches and decision makers to define successful strategies and the need of analyzing large sport databases. The aim of our topic is to show how Principal component analysis of distributions is be very helpful in sport analytic. To do this, we consider two databases on soccer and basketball players from the Sports Analytic R package. These data contain the individual performances of players. By considering teams instead of players, as the unit of analyses, we determine then the distributions of these teams. To do this, we construct an empirical distributions tables of teams via our R package graphPCA. After that, we apply a principal component analysis method to our table of distributions. In our application, we show how the application of PCA to the distributions is very helpful for data visualization and decision making.