Characteristic Analysis and Prediction of Traffic Accidents in the Multiethnic Plateau Mountain Area

Abstract

This paper analyzes the distribution characteristics and influencing factors of traffic accidents in the multiethnic-gathering plateau mountain area. Using data from an investigation of 56,651 accident records in Guiyang, China, in 2015, we analyze the spatial and temporal distribution characteristics of accidents, and build a Gray Markov model to predict regional traffic accidents. Accident types of rear-end collisions and yielding not according to regulations are the most common ones in Guiyang, and the majority of these types of accidents occur along main roads and near commercial areas. Meanwhile, logistic regression analyses are applied to determine the correlations between road type, traffic flow type, and traffic accidents. It shows that road type and traffic flow type have no significant interactive impacts on accidents, while road type has a significant impact on the number of accidents. The accident distribution is more intensive in the northern part of the city where many ethnic groups gather together than others. Therefore, it is recommended that the traffic management department consider ethnic differences when improving traffic safety in this area.