Research on Hotspot Identification Method of Urban Road Traffic Accidents Based on Improved Kernel Density Clustering Algorithm

Based on the limitations of the existing methods for identifying urban road traffic accident hotspots, a DENCLUE clustering algorithm based on Bayesian optimisation is proposed to identify accident hotspot areas. Using the real application cases in S district of Urumqi city, it is shown that the improved DENCLUE clustering algorithm not only avoids the influence of manual parameter adjustment on the results, but also can comprehensively consider the spatial effect of the accident point on the neighbouring area, abandons the traditional practice of predefined division of the target area, and is able to identify clusters with arbitrary shapes, which effectively solves the defects of the current identification technology and provides a new method for the exploration of the spatial distribution characteristics of the accident prone This effectively solves the shortcomings of the current recognition technology, and provides an effective means to explore the spatial distribution characteristics of accident-prone locations.