Ryan Parker, College of Charleston


Modeling Basketball's Points per Possession

Abstract:In this problem we consider how to model the number of points a basketball team scores on a possession, where we use combinations of players as the predictors of the number of points scored on an individual possession. We illustrate how Normal, Poisson, Negative Binomial, Zero Altered Poisson, and Multinomial logit regressions model this data, and we show that the Multinomial logit regression is the best at modeling this data.

Mentor: Amy Langville and Martin Jones (College of Charleston)