CONTACT US
AddressFaculty Wing F416, Munjiro 193, Yuseonggu, Daejeon, 34051, South Korea
Tel+82 42 350 1284
Emailtops@kaist.ac.kr
Identifying hazardous locations on highways is an essential step for safety improvement programs and projects since it provides decision makers with a logical and scientific basis for the allocation of resources in a cost-effective manner. There have been numerous studies conducted to develop suitable methodologies for identifying hazardous locations; however, most of them have not considered spatial interactions which are inherent in traffic accidents. In this paper, we use the GIS-based geographically weighted regression (GWR) that can model crash outcomes and identify hazardous locations on expressways while reflecting the effect of spatial dependency and heterogeneity on the outbreak of traffic accidents. This method has been applied to a case study at Gyeongbu Expressway in Korea with 3-year crash data. Koenker's studentized Bruesch-Pagan and Moran’s I tests confirm the spatial relationship among crash data. The findings indicate that it is proper to model crash frequency with GWR for identifying hazardous locations.