Predicting Visitor Distribution for Large Events in Smart Cities

The prediction of the distribution of visitors in large events is a valuable piece of information in the context of smart cities. The organizers of large events leverage it for safety and coordination purposes and the Fog computing infrastructures for cost effective, agile and reliable allocation of the mobile apps and festival services workload along the continuum from edge devices to cloud. In this research we examine two sets of supervised Machine Learning techniques in order to predict the visitors' distribution in the next timesteps and evaluate them using real data from a large music event that took place in 2017 and 2018. To enrich the feature space of the predictive models we use and evaluate open data such as the weather and the popularity of artists. A further added value of the examined Machine Learning techniques, in comparison with the current state of the art in mobility prediction, is that they look into the phenomenon of visitors coming and going from the area of interest.