Macroscopic Patterns in Sparse Location Data

Lüders, Ruppel – 2018 – Macroscopic Patterns in Sparse Location Data Identifying Mobility Prototypes

Ubiquitous computing and location-based services are key enablers for gaining novel insights into human movement behavior on a large scale. When searching for patterns in movement behavior, several approaches have been proposed to compare location trajectories or semantic overlaps. This talk and paper introduces the notion of mobility prototypes, which describe human mobility behavior at a macroscopic level and express its most significant characteristics. Based on a dataset of coarse
spatiotemporal data from almost 4000 mobile users of a locationbased service over a period of 14 days, we outline a comprehensive and systematic pipeline for processing and exploiting mobility traces and provide insights into the mobility behavior in the city of Berlin. We present a framework to characterize
individual users precisely by employing a novel combination of discriminative features capturing the macroscopic mobility behavior, without revealing their identity or actual whereabouts, and define mobility prototypes by constructing clusters of users. The results yield valuable insights into the variety of human mobility and open up new possibilities for future mobile services.