Document Type : Research Paper
Authors
1
Department of civil engineering ShQ.C,Islamic azad university,Shahr-e-Qods,Iran
2
Department of Civil Engineering, Tehran Marqaz Branch, Islamic Azad University, Tehran, Iran
3
Department of Civil Engineering, Shahid Beheshti University, Tehran, Iran.
4
Department of Civil Engineering, Babol Branch, Islamic Azad University, Babol, Iran
Abstract
This study presents a comprehensive spatial analysis of traffic accidents on Iran’s road network from 12019 to 2023, utilizing advanced machine learning techniques to identify and classify accident hotspots. The research employs a two-stage clustering methodology: DBSCAN (Density-Based Spatial Clustering of Applications with Noise) for geographic hotspot identification, followed by K-Means clustering for risk stratification. Analysis of 82,157 validated accident records revealed 923 significant hotspots, with 16 critical risk locations, 440 high-risk zones, and 467 moderate-risk areas. The study incorporates temporal features, weather conditions, road characteristics, and collision types to provide a multi-dimensional understanding of accident patterns. Results indicate that approximately 21.6% of clustered accidents occur on holidays, and 9.3% occur under adverse weather conditions. The top critical hotspot, the Farouj-Shirvan corridor, recorded 239 accidents with an EPDO score of 1,334, predominantly involving rollover collisions. The findings demonstrate the effectiveness of density-based spatial clustering combined with multi-dimensional risk assessment in identifying high-priority locations for traffic safety interventions on Iran’s road network.
Keywords
