Data analysis of basketball game performance based on bivariate poisson regression model

Ke Shen

COMPUTER MODELLING & NEW TECHNOLOGIES 2014 18(12C) 474-479

Physical Education Institute, Hunan University of Technology, Zhuzhou 412007, Hunan, China

Conventional methods used to process two-dimensional discrete data will produce large errors and have a narrow scope of application. Along with the development of mathematical theories and computer technologies, some scholars propose to process two-dimensional discrete data by bivariate Poisson regression model, which takes correlation among data sets into consideration and has excessive variability so that results of data analysis can be more accurate. This paper firstly introduces bivariate Poisson distribution and bivariate Poisson regression model, and then uses this model to analyze performance data of each team in regular seasons of 2013-2014 CBA (China Basketball Association) and 2012-2013 NBA (National Basketball Association), and predict performance in post seasons. Through comparison between actual results and results of double independent Poisson distribution, this model can better predict game performance.