Traditional factors analysis of accidents is usually based on full sample analysis, which cannot reflect the difference between different types of responsible parties. In order to explore such differences in order to the accurate causal analysis results, based on the accident data of Yunnan Yuan (Mou) Shuang (Bai) secondary highway from 2012 to 2017, the latent class model was adopted to classify the responsible parties according to the age, gender, mean of transportation and cause of the accident. Then, the accident severity analysis model of various groups of perpetrators based on the Logistic model was built, with whether the accident was working day, the accident time, the mean of transportation of the non-responsible party and whether the intersection was in 100m before and after the accident site as independent variables, and the accident severity was taken as dependent variable. The results show that the traffic accidents can be divided into three categories from the perspective of the responsible parties. The severity analysis of different kinds of accidents shows that the influence of different factors on the severity of different kinds of accidents is quite different. The results are helpful to understand the factors of various traffic accidents and provide a new view for the traffic safety measures of secondary highways in mountainous areas.

CICTP