Investigating the relationship between traffic incidents and public events: A case study

Large social events can influence traffic conditions and possibly lead to jams and incidents. This study leverages crowdsourced data to analytically evaluate the relationship between social events and traffic incidents in the city of Chicago. In particular, we collected data on social events from scraping online webpages, as well as traffic data from a twitter account that posted irregular traffic incidents based on a crowdsourced navigation application (Waze). Using these two sources the relationship between social events and the occurrence of traffic incidents was investigated. The total number social events and their categories for each region and its neighboring regions were used to build models that predicted the chance of a traffic incident occurrence. Based on the analysis, we demonstrated the variables that indicated significant influence on the chance of a traffic incident occurrence in the same day, such as the number of festivals and fairs, total number of events in the (neighboring) region. We have also developed and tested several models to predict the traffic incidents. The results indicated that using solely online listed social events may not be sufficient for traffic prediction. Although the accuracies are not considerable for an independent model, it clearly indicates that the additional information provided by social events will be a valuable addition to the existing traffic prediction models.