Development of a traffic accident simulation system for main roads in Loei Province, Thailand: Application of a geographic information system and multiple logistic regression with clustering

Abstract

Traffic accidents are a major and crucial problem worldwide. The development of a traffic accident simulation system applied by using a geographic information system and multiple logistic regression with clustering can provide drivers with safe routes as well as guidelines for assessing the risk points of accidents in each subdistrict. This research is based on case-control study design. The data were collected by using two types of questionnaires: a questionnaire for 35 community leaders and a questionnaire for 580 community residents based on the distance at which main routes pass through the subdistrict area. The data were analysed through multiple logistic regression with clustering, and the standardized coefficient of the selected variables was then added to the equation as a weight in the traffic accident simulation system. The results of the study indicated that 11 variables affected traffic accidents. These factors were evaluated in order to predict traffic accidents (Pseudo R square=0.5906). Standardized coefficient of variables was applied in a geographic information system to simulate traffic accidents on roads. This study was distinctive for its analysis, which examined the clusters of variables that were the subdistrict-level data, including surroundings and road conditions at the riskiest location in each subdistrict. The data were analysed based on their quality as subdistrict data clusters. The analysis results were then applied as the weight of variables used in the GIS to obtain the values appropriate to the data clusters’ quality for the GIS to properly simulate traffic accidents in each area.