Examining Heterogeneity of Driver Behaviour Using Temporal and Spatial Factors

摘要

Temporal and spatial characteristics of the road environment are known to influence driver behaviour andconsequently the risk of an injury or fatality crash. Nonetheless, much of our understanding of the risksof injury and fatality associated with driving relies heavily on police crash records. These capture themost serious of crashes but underreport other events. Studies which rely on these data sources typicallyignore the temporal and spatial factors. Advances in technology have enabled more detailed study ofdriving on a day-to-day basis and therefore the opportunity to examine driver behaviour for the samedriver across time and space. However, this has brought with it its own challenges. This includesextensive intra and inter-driver heterogeneity which is not apparent when using ‘traditional’ datacollection methods. This paper presents a framework and methodology for isolating the influence ofdrivers’ inherent characteristics on driver behaviour. This is done by constructing temporal and spatialidentifiers which control for the influence of the road environment. Results of analyses conducted usingempirical driving information collected from 106 vehicles in Sydney, Australia to examine theeffectiveness of this approach are included. The results indicate that in 80 percent of road environmentsthere is less intra-driver variability in speeding behaviour than inter-driver variability when accounting fortemporal and spatial characteristics. Clustering and regression analyses for the most frequently observedroad environments are also presented. Driver personality characteristics are significant for evening tripshome on residential roads and acceleration profiles significant in evening trips on roads with 50 and 60km/h speed limits.