Unveiling the Dynamic of Traffic-Crowding Relation Through Mobile Phone Big Data Analysis

This study investigates the potential of mobile phone data for traffic flow estimation in urban areas. More in detail, we use three distinct sources of mobile phone data to monitor traffic dynamics between pairs of three neighboring small areas—three “Aree di Censimento” (ACE)—within the Province of Brescia. Two indicators designed to capture crowding and traffic intensity are defined, and their relation and influence on traffic flows are investigated. Employing a vector autoregressive model with exogenous variables, which includes complex seasonality through dynamic harmonic components and the two indicators defined on mobile phone data, we assess its efficacy in comparison to a baseline model without these indicators. We find that the proposed model consistently provides satisfactory traffic flow estimates and improves the estimation of traffic flows with respect to the baseline one. More in detail, we observe that the traffic intensity indicator significantly impacts traffic flows, while the crowding indicator, although significant, exerts a negative influence. Despite moderate variations in the impact of seasonal components among ACE pairs, this research underscores the potential of integrating mobile phone data into advanced modeling frameworks for traffic monitoring