Cause Analysis of Railway Traffic Accidents Based on Random Forest

Abstract With the rapid growth of the railway operation scale, all kinds of railway traffic accidents happen from time to time, so it is great significant to accurately identify the main influencing factors and their influence degree. In this paper, random forest model is proposed to analyze the cause of railway traffic accidents. Considering people, equipment, environment and other aspects, 11 influencing factors were extracted from 491 accident data. The influence degree of different factors on the severity of the accident is judged through variable importance measures of the random forest. On the basis of this, some suggestions are put forward for raising the safety level of railway transportation. The results show that the random forest model is accurate for analyzing the causes of railway traffic accidents, which can provide decision support for railway transportation safety management.