Detection of application-relevant user groups in anonymised in-app location data

Location data collected by mobile applications is typically aggregated from a heterogeneous
sample of mobile devices with varying demographic and behavioral characteristics. In this
paper, we present a method for detecting homogenous groups of mobile devices relevant to
specific scientific domains. We apply this method to an anonymized in-app location dataset of
~2,000,000 mobile devices in the United Kingdom, to detect homogeneous groups of devices
relevant to applications including transport planning and infectious disease modeling.