Using multi-regression to analyze and predict road traffic safety level in China

Abstract

Analyzing and predicting road traffic accident severity is of vital importance for improving country road safety management capacity. Investigation of relationship model between road traffic deaths and factors reflecting traffic development and management level enables managers to carry out calculations in order to predict and analyze the macro trend of road traffic safety. Therefore, this paper deals with a multiregression model to predict road traffic fatalities and examine the significant factors of road traffic safety. In this paper, the data from 2002–2012 Annual Statistics of Road Traffic Accidents in China (ASRTAC) are collected to model the correlation of road traffic fatalities with a set of factors (i.e. vehicle park, roadway lane-miles, total number of licensed drivers, Gross Domestic Product, number of traffic fines and number of serious traffic accident), which in return, the proposed model can be used to predict the road traffic deaths in future years. In this work, we show how the Multi-Regression (MR) method is performed to predict the road traffic accident deaths with a set of related factors. Then, hypothesis testing is used to examine the significance of the proposed model and each item. In addition, the backward strategy was used to remove some insignificant items and modify the prediction model. Finally, the improved model can be applied to predict the traffic fatalities and its confidence interval. The results show that the recent road traffic deaths in China have significant linear relationship with GDP, the number of traffic fines and the number of serious traffic accidents, and the proposed model can provide an accurate traffic deaths prediction. In general, this paper contributes to prediction of the road traffic fatalities, which provides the traffic management administration guidance for policy development and resource allocation regarding road safety in China.