Neural networks in football

Abstract

Machine learning provides the ability to examine massive datasets and discover patterns within the data without relying on a priori assumptions. Its application to the field of sport (which is experiencing rapid growth) is divided into predictive (training programmes, results...) and explanatory (injuries) models. In this report, which is part of a final project for a master’s degree course, we use unsupervised learning techniques (self-organised maps and clustering) to group players according to different statistics (passes, goals, fouls, etc.) and compare the results with their real playing positions. We also describe the tools used to implement and visualise the results, so that a reader can be inspired to carry out their own project.

Keywords

• mapas autoorganizados
• shiny
• r
• fútbol
• inteligencia artificial
• aprendizaje no supervisado