search for


Analysis of K-league data using bivariate Poisson and diagonal inflated model
Journal of the Korean Data & Information Science Society 2018;29:1643-53
Published online November 30, 2018
© 2018 Korean Data and Information Science Society.

Yun Seo Heo1 · Kyoung Hee Kim2

12Department of Statistics, Sungshin Women’s University
Correspondence to: Assistant professor, Department of Statistics, College of Natural Sciences, Sungshin Women’s University, Bomun-ro 34 Da Gil, Seongbuk-Gu, Seoul 02844, Korea. E-mail:
This work was supported by the Sungshin University Research Grant of 2018.
Received October 29, 2018; Revised November 16, 2018; Accepted November 16, 2018.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License ( which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
There has been a steady research for analyzing number of goals of soccer game (Seong and Chang, 2007; Lee, 2012). In this study, eight regression models including the bivariate zero inflated Poisson regression model were fitted to K-league data for 2015-2018. The response variable is the number of total goals or the second half goals for home and away teams. Explanatory variables are the number of goals for the first half and the first half ball possession rate of each team. We chose bivariate Poisson regression model and the diagonal inflated regression model with Poisson tie probability distribution following several model selection criteria such as log likelihood, AIC and BIC. We found that the first half goals of home teams have a higher influence on total goals of home teams than the first half ball possession rate, but vice versa for away teams.
Keywords : Bivariate zero-inflated poisson, diagonal-inflated, K-league.