Citation

Tamakloe R, Park D. Int. J. Urban Sci. 2023; 27(3): 483-517.

Copyright

(Copyright © 2023, University of Seoul, Publisher Informa - Taylor and Francis Group)

DOI

10.1080/12265934.2022.2134182

PMID

unavailable

Abstract

Studies have employed several techniques to identify the effect of individual risk factors influencing crashes at hotspot locations. Nevertheless, as crashes are sometimes influenced by a combination of risk factors, identifying the chains of factors collectively contributing to fatal crashes at hotspot locations could provide added insights for improving traffic safety. By employing fatal crash data from Korea, this study identifies hotspots with increasing (critical) and decreasing (diminishing) temporal trends using a spatio-temporal hotspot analysis tool in GIS. Further, a machine learning technique is employed to explore the chains of factors influencing the number of vehicles and the number of casualties involved in fatal crashes at intersections and midblocks in each hotspot type identified. In general, results showed that minibuses/vans and construction vehicles were mainly at fault for fatal single-vehicle pedestrian-involved crashes. While many casualties and vehicles are likely to be involved in crashes at midblocks during the daytime regardless of the hotspot type, the nighttime variable was particularly associated with large casualty-size crashes at critical intersection hotspots. Further, while reckless driving was mostly associated with single-vehicle crashes at intersections in diminishing hotspots, pedestrian protection, and improper centreline crossing violations were more pronounced at midblocks in diminishing hotspots. This analysis identified groups of factors that could be collectively controlled to improve road safety and proposed countermeasures to mitigate fatal crashes on roadways.

Language: en

Keywords

fatal crash; intersection; machine learning; pedestrian; Spatio-temporal hotspot analysis; temporal trends