Identifying Interrelated Factors of Fatal and Injury Traffic Accidents Using Association Rules

Abstract

This study aims to investigate the possible relationships of risk factors related to traffic accidents playing important roles in increasing the likelihood of accidents. In the previous studies, parametric models are mostly used to investigate the causes of traffic accidents. As a non-parametric data mining model with its increasing usage in recent years; association rule mining was employed in this study to analyse the traffic accident data for the period of 2015 and 2020 in the city of Sakarya, Turkey. The analysis of the data studied revealed the relationships among the external/environmental, driver, road, vehicle and nature of accident factors. Some important rules regarding accidents occurring on daylight came into prominence within the scope of this study. In addition, the correlations between the driver casualties and their education level and ages are established to be related. The findings are beneficial for transportation authorities to apply effective operational strategies and campaigns to increase the road safety.