Estimating traffic flows from vehicle trajectories based on sparse mobile phone geolocation data

Other conference contribution, 2025

Empirical traffic flow data are key in transport, mobility, and spatial planning. They can help understand congestion, traffic safety, environmental exposure and risks, among others. At scale, they can support data-driven decision making, helping to decide where interventions are most needed.

However, existing traffic flow data from sensors or traffic counts [1] lack spatio-temporal coverage and granularity. Other data, e.g. from navigation API’s, are proprietary, commercial or limited-access, and unavailable to decision-makers. Large mobile phone traces data recently emerged as a promising source to capture dynamics at scale given their size, granularity, and coverage. They have been used to analyse travel demand (origin-
destination), activity-locations, and individuals’ activity spaces. Yet, despite their potential for exploring trajectories and traffic flows [1], dynamic applications other than understanding pedestrian routing behaviour [2] remain unexplored.

This study aims to explore how traffic flows with high spatial and temporal coverage and granularity can be estimated from vehicle trajectories based on sparse mobile phone geolocation data. We develop a methodology to create trajectories and flows from raw location data and test how various parameters affect the results. We contribute our methodology, code and data to allow for replication in other studies, and reflect on directions for future development.