Figure 3.

Comparing football match attendance figures to mobile phone and Twitter activity. (a) We investigate whether there is a relationship between the number of people attending each football match and the recorded mobile phone call and SMS activity inside the stadium. We find a linear relationship between these two variables (adjusted R²=0.771, N=10, p<0.001, ordinary least-squares regression). (b) Similarly, we find a pattern consistent with a linear relationship between Internet connection activity in the stadium and the number of attendees at each match (adjusted R²=0.937, N=10, p<0.001, ordinary least-squares regression). (c) We also observe a linear relationship between Twitter activity in the stadium and the number of match attendees (adjusted R²=0.855, N=10, p<0.001, ordinary least-squares regression). (d) We explore whether this relationship could be exploited to infer the number of attendees from communication data if no other measurements were available. Using data on Internet activity, we build a linear regression model using only nine out of the 10 football matches and then predict the attendance at the 10th match. We then repeat this leaving a different match out every time. Here, we plot the resulting estimates and their 95% prediction intervals. We find that the actual number of attendees falls within the 95% prediction interval for all 10 matches.