Identifying Blackspot Sections of Accident Using Clustering Algorithms

Articles in Press

Document Type : Research Paper

Authors
Mona Delshad 1
Hani Rezayan 2
Javad Sadidi 3
Akbar Danesh 4
1 M. Sc of Remote Sensing and Geographical Information System, Kharazmi University, Tehran
2 Assistant Professor, Department of Remote Sensing and Geographical Information System, Kharazmi University, Tehran
3 Associate Professor, Department of Remote Sensing and Geographical Information System, Kharazmi University, Tehran
4 PhD in Civil Engineering, Road and Transportation, Tehran
10.22119/jte.2025.495110.2724
Abstract
Identifying road accident hotspots enables better understanding of accident patterns for road safety experts to improve roads safety. This study as an applied research, analyzes road accidents to identify blackspots using clustering algorithms: Kernel Density Estimation (KDE), Local Moran's I, and K-Means. The Iranian Ministry of Roads and Urban Development's definition of blackspots was used as the reference for algorithms comparison and results validation. The study examined road accident data on the Mashhad-Bojnord road from 2019-2022. After extracting blackspots using the mentioned algorithms in raster format, their overlap with the reference blackspots was calculated. The KDE algorithm showed the highest match with the blackspots, with 82.55% and 70.96% overlap in forward and return directions, respectively. Analysis of blackspots and accidents distributions near settlements revealed that the highest accident density occurs in the first 30 kilometers of Mashhad city, between Farooj and Shirvan cities, 14 kilometers after Shirvan city exit in the forward direction, and the first 8 kilometers of Bojnord city.

Keywords

Subjects

Quarterly Journal of Transportation Engineering

Articles in Press, Accepted Manuscript
Available Online from 27 July 2025

  • Receive Date 22 December 2024
  • Revise Date 08 June 2025
  • Accept Date 13 July 2025