Models valuing actions in football
A key model in evaluating football players is called expected threat. It is a machine learning method, where the value of an action – a pass, a dribble etc. – on the field is assigned a value based on how much it increases the probability of scoring.
In this project, the students will work with company Twelve football to develop a general tool for fitting expected threat models. They will work with data from professional football — both men and women's leagues — and implement regression methods. The students should be willing to learn how to use Streamlit as a tool for producing an interactive ML platform. They will be supervised and work within Twelve football and can, if they want sit at their offices at Snickeriet in Stockholm.