Two-dimensional Regression Modelling with Copula Dependencies and a Focus on Count Data and Sports Applications (English shop)

Preview

Extract, PDF (380 KB)
Table of Contents, PDF (93 KB)

This work was fundamentally motivated by the application of regression modelling to football match outcomes and specifically FIFA World Cups. Modelling football results is a rather popular topic. However, there is no gold standard on how to tackle the bivariate nature of match outcomes and how to assess the possible underlying dependency thereof. This work proposes to use the mathematical structure of copulas to include the bivariate dependency structure within the modelling process. In this context, two regularization approaches were implemented into existing infrastructure and multiple showcases of the methodology are presented.