2025 年 16 巻 論文ID: PP4173
This study aims to quantify accident risk in relation to road environment factors. In addition to road structure, topography, population distribution, and facility location, which have been used as road environment factors in previous studies, this study also considered the road network structure. Using a neural network, accident risk was estimated from road environment factors in the central area of Tottori City, Tottori Prefecture, Japan. Based on the estimation results, the dangerous locations were identified in the target area. In addition, LassoNet was used to identify road environmental factors that strongly influence accident risk prediction. This study demonstrated the effectiveness of the identified road environmental factors in reducing accident risk.