Social Big Data can predict the behaviour of football fans
”I would bring my daughter ❤❤she is totally bitten with football”, ”Hopefully we won't underestimate the Irish”, ”I'd be happy if Fischer/Mainz05 enters the field”.
These are comments posted on the Facebook page of the men's Danish football team ahead of the match against Ireland on Saturday 11 November. According to a new research project, the activity on this page bodes well for the support for this match.
Facebook is a community in which football fans can express their feelings and get excited about football together. This makes Facebook very interesting to Social Big Data analysts like Professor and Director of Centre for Business Data Analytics, CBS, Ravi Vatrapu at Copenhagen Business School.
His team of researchers and students from CBS has developed a model to predict the behaviour of football fans. By monitoring the activity on the Facebook page of the men's national football team between 2014-2016, the research team demonstrates a connection between the number of posts, comments, likes ahead of a match and the size of the audience attending the match.
”Big social data from Facebook can almost precisely predict the amount of spectators for a football match. Our study shows that the more football fans who are active on the fan page the more people will either show up for the match or watch it on TV,” says Ravi Vatrapu and continues:
”This is worth a lot for companies to understand their customers and how to improve marketing strategies. ”
More Facebook activity, more tickets sold
The research builds on existing work on using big social data for predictive modelling. However, the study adds a new domain; sports. Not many have used Facebook data for this field. Twitter and Google Trends have been more used for these prediction studies, for instance predicting movie revenues and iPhone sale from tweets, says Ravi Vatrapu.
The owners of the fan page, the Danish national stadium Parken and the Danish Football Association (DBU) can gain a lot of relevant input from this activity. Ravi Vatrapu strongly recommends that DBU orchestrates more activity on this Facebook page as it will ensure more viewers and tickets sold. And as a piece of bonus information for DBU, the study has shown that the initiatives that activate the users the most are pictures, not videos.
”Photos result in more involvement than videos, so DBU will get more bang for the buck if they spend their marketing budget on photos. Their followers do not take the time to watch the videos,” says Ravi Vatrapu.
The study does not show why followers do not want to spend time watching the videos, but research is also relevant for other elements in Parken's overall business.
”When big social media data can predict the number of spectators, it is of great value to Parken. They will be able to tell advertisers in advance more precisely how many spectators they will reach, how many police officers they should book for a match, or how much beer to keep in stock.”
Prediction model also valuable to other industries
The future perspective for this kind of research is not uninteresting. In terms of adapting the prediction model to other contexts, the possibilities are endless, according to Professor Ravi Vatrapu.
”Public health could be analysed. For instance, we could analyse the development of obesity and diabetes in relation to the activity on the Facebook pages of McDonalds and Coca-Cola,” says Ravi Vatrapu.
Read the research publication ”Big social data analytics on football: Predicting spectactors and TV ratings from Facebook data”
The research was published at the IEEE Big Data Congress 2017 Conference in Honolulu this summer with a peer review process prior to admission.
For more information, please contact:
Professor at CBS Ravi Vatrapu
Journalist at CBS Matilde Hørmand-Pallesen
- The research team included three students; Nicolai H. Egebjerg, Niklas Hedegaard and Gerda Kuum and two researchers from Centre for Business Data Analytics, CBS; Associate Professor Raghava Rao Mukkamala and Professor Ravi Vatrapu.
- The analysis is based on activity on the Facebook page of the men's national team between 30 October 2014 and 10 November 2016, which covers the 11 international matches played at home.
- The researchers registered the number of posts, comments and likes ahead of each game and compared them with data on match dates, the number of spectators per match and TV ratings.
- This study shows that traditional sports analytics measures of Game Outcome Uncertainity and Quality of Players are not necessary when using big social data.
- The model demonstrating the predictions builds on GLM code.