About · Gregory Everett Link to heading

About Me Link to heading

I am Gregory Everett, a third-year PhD student in Computer Science at the University of Southampton. I also completed an Integrated MEng in Computer Science at the University of Southampton, earning a first-class degree. During my studies, I completed a variety of modules related to machine learning.

My PhD research is focused on optimizing and analyzing teams through machine learning methodologies such as Reinforcement Learning and Supervised Learning models like Graph Neural Networks. I concentrate particularly on football teams, aiming to advance research into how data science can improve decision-making processes in clubs for recruitment, tactical analysis and player evaluation.

I am passionate about interdisciplinary collaboration within both academia and industry, and I am committed to helping others apply and understand machine learning within their fields. I believe that machine learning is most powerful when combined with human domain knowledge, especially from football experts. I have collaborated with sports data and analytics companies, leading to publications in top academic and industrial conferences. Additionally, I have worked alongside researchers in various AI domains, including robotics and interpretable machine learning, further broadening my research scope and impact.

I also have experience working in the sports industry, having completed an internship at Sentient Sports looking at applications of AI for sports prediction and player evaluation. I have also collaborated closely with the company throughout my PhD. Additionally, I am involved in undergraduate teaching - running lab sessions for modules such as Intelligent Agents.

Research Interests Link to heading

My doctoral research primarily focuses on AI-driven prediction and optimization in football. With the recent rise in spatial data, including event and tracking data, the field of data science related to performance evaluation and optimization has grown significantly in football. This evolution has provided clubs with opportunities to evaluate players and strategies in greater detail using novel performance metrics and prediction methods.

I extend this research by modeling football teams as multi-agent systems and applying these concepts to a variety of tasks. These include:

• Measurement and optimization of spatial teamwork: Utilizing positional data to analyze player interactions and how these can effectively reduce opponent threat.
• Player location prediction: Developing models to estimate tracking data from event data, thereby making tracking insights accessible to clubs and researchers without direct access to tracking data.
• Team selection and substitution optimization: Implementing methodologies based on Reinforcement Learning to maximize team performance while enhancing player welfare.

Through these research initiatives, I aim to advance the application of data science in football, ultimately contributing to improved performance evaluation and strategic decision-making within the sport.