Analysing Spatial Intrapersonal Variability of Road Users Using Point-to-Point Sensor Data

The availability of newly emerging forms of data in recent years has provided new opportunities to study spatial intrapersonal variability, namely the variability in an individual’s destination and route choices from day to day. As well as providing insights into traveller needs, preferences and adaptive capacity, spatial intrapersonal variability can also inform the development of user classes for models of network disruption and for measuring behaviour change to evaluate the impact of network changes. This paper proposes a methodology for measuring spatial intrapersonal variability using point-to-point sensor data such as Bluetooth or number plate data. The method is innovative in accounting for sensor specific probabilities of detecting a passing device or vehicle and in providing a single measure for each traveller which considers destination and route choice variability and both the quantity of different trajectories utilised as well as the intensity with which they are used. A data science method is also presented for examining relationships between different trajectories observed in the network based on whether they are typically made by the same travellers. A case study using 12 months of real-world data is presented. The example provided demonstrates that a substantial amount of data processing is required, but the outputs of the methods are easily interpretable. Perhaps surprisingly, the analysis showed that the trips people made on weekdays were more evenly spread across a range of different trajectories than the trips they made during the weekend which were more concentrated into a few spatially similar clusters.