| Title |
A Machine Learning?Based Decision Support Model for Selecting Pedestrian-Priority Road Candidate Segments - A Link-Level Probabilistic Approach and Explainable AI Analysis |
| Authors |
이아진(Lee, Ah-Jin) ; 최강록(Choi, Kang-Rok) ; 공은미(Kong, Eun-Mi) ; 박훈태(Park, Hoon-Tae) |
| DOI |
https://doi.org/10.38195/judik.2026.08.27.4.137 |
| Keywords |
보행자우선도로; 기계학습; CatBoost; SHAP 분석; 의사결정 지원 Pedestrian-priority roads; Machine learning; CatBoost; SHAP analysis; Decision support |
| Abstract |
Pedestrian-priority roads are designated on shared streets under the Promotion of Pedestrian Safety and Convenience Enhancement Act to improve pedestrian safety and walking conditions. However, current designation criteria have limitations in quantitative comparison and prioritization. This study developed a machine learning? based classification model that learns the spatial and environmental characteristics of existing pedestrian-priority roads and identifies high-priority candidate segments among nationwide local streets. Using 29 spatial variables, the model estimated link-level designation probabilities and achieved a recall of 93%. SHAP-based analysis was conducted to identify major influencing factors, and link-level results were aggregated into street-name units to examine their applicability to administrative practice. The proposed approach provides a data-driven framework to support policy decision-making for pedestrian-priority road designation. |