국내 철도 사상사고 분석 모형 연구

초록

The railway casualities are rarely occurred so that the railway accident modeling needs to consider its low casuality. This study aims to investigate factors influencing injuries and fatalities from railway accident record data collected for 3 years from 2014 to 2016. Many cases of the railway accidents with no injuries or fatalities were recorded. We developed the zero-inflated Poisson models to consider the excessive number of zero injury and fatality on many railway accidents. The zero-inflated Poisson models of casualities and fatalities investigate characteristics of railway accident such as train type, organization, accident factor, and accident type. Then we addressed the model performance by comparing with the general Poisson model. The results show that the zero-inflated Poisson regression model of casualities is superior to general Poisson regression model, while the model for fatalities has no superiority against general one. The results of this study indicate that the railway accident models of casualities and fatalities should be separately developed.

키워드