Improvement of occurrence rate of railroad accident using hierarchical cluster analysis

To improve the occurrence and fatality rates of railroad accidents, it is necessary to promote railroad safety improvement projects continuously. However, in the current guidelines for economic analysis of the projects, it is difficult to sufficiently reflect the effect of preventing railroad accidents of railroad safety improvement projects, and this has led to limitations in actively promoting railroad safety improvement projects. The accident rates in the guidelines divided by the type of railroads (high-speed railroad, inter-city railroad, conventional railroad, urban railroad) make it difficult to represent actual accident rates, and they also do not reflect the effects of recent railroad accidents. Therefore, the benefit of railroad accident reduction based on the accident rates within the guidelines may be less accurate. This study aims to make more rational investment decisions by proposing a process that can continuously update and calculate railroad accident rates. To this end, it is confirmed that the accident rates in the current guidelines are unrealistic, and a classification that can achieve representativeness by railroad type is derived through hierarchical cluster analysis. In addition, by using the latest railroad accident data, the effects of recent railroad accidents are updated to present realistic benefit results. Comparing the results of the process proposed in this study with existing guidelines emphasizes the importance of continuous improvement of accident rates using this process.