Modified-DBSCAN Clustering for Identifying Traffic Accident Prone Locations

摘要

Road traffic accidents, especially expressway traffic accidents, have become a severe problem in China. Under this condition, identification of road traffic accident prone locations is in urgent need. This work proposes a modification of DBSCAN clustering algorithm with parameters ε and minPts carefully chosen for identifying traffic accident prone locations. Experimental results on traffic accident datasets of three national expressways are given, demonstrating the effectiveness of the proposed algorithm.