Skip to main content Skip to Footer

May 06, 2016
Guest Blog Post – UCL & Accenture Analytics Pre-Match Predictions Project
By: Zhaozhi

UCL Logo

Hi, I’m Zhaozhi a student at UCL. I’m pleased to write a guest post on the Accenture Graduate Blog to tell you all about a project I worked on in partnership with Accenture Analytics and my university department.

Firstly, a bit about me. I started at UCL in 2015 and I’m currently studying towards an MSc Computational Statistics and Machine Learning. In the future, I hope to become a data scientist, as I enjoy finding actionable insights from data.

The project I was involved with was a real-world data problem set by my department in partnership with Accenture. I worked with my UCL supervisor Franz on creating models that would enable us to predict the winner of each match in the RBS 6 Nations Championship. I had to systematically study all of the pre-match predictions for the outcomes of matches, as well as using in-game models. The data set I was working with (Accenture provided the data which is owned by Opta) was very detailed. Being able to work with this data was fascinating because it was 15 years’ worth of historical matches and captured almost all events happening on the field every second, both at team and player level. I then built a problem specific model evaluation framework, which would allow me to compare the models. Using the evaluation framework, I was able to explore and analyse the effectiveness of the different prediction models.

Each week my supervisor and I would have a call with our Accenture contact Nina Fertacz (Accenture Analytics Senior Manager), where I would summarise my progress. It was initially quite tricky to present quite complex and abstract concepts without using any visual graphics or presentation slides! However, my presentation skills definitely improved as a result of this. I also really enjoyed discussing the problems – they really got me thinking about the complexity of the problem and I liked being able to talk things through with experts in the field of analytics.

Overall, I thoroughly enjoyed working on this project with Accenture. Although aspects of the project were challenging due to the fact that final outcomes in competitive sports can be determined by a number of different factors, it was a great opportunity to learn from experts in the industry and test my knowledge of analytics. The project was relatively short (two months), which meant although I made improvements to the model, it was still quite simple and there are various aspects that I would have liked to investigate further if time had allowed me to. I am looking forward to the possibility of further collaborations in the future!

Analysis Team

The Accenture Analysis team

More blogs on this topic