Prediction Model for Severity and Number of Accidents at Weaving Section (Roundabout) Based on Geometric Characteristics using Logit-Based Analysis and Gene Expression Algorithm (Case Study: Qom City)

Roundabouts usually have higher traffic safety compared to other types of intersections. Despite the transformation of intersections into roundabouts, high-severity accidents still occur in them. Improving accident prediction models and their parameters can help prioritize action and reduce the cost of geometric modification. This research aims to investigate the geometric factors of severity and the number of accidents in the weaving section at these roundabouts. In this research, by using the accident statistics of six roundabouts located in Qom city, the influencing factors on the severity and number of accidents were investigated using the logit model and gene expression algorithm. The results showed that the width of the weaving section had the greatest impact on the occurrence of accidents compared to other geometrical variables of the roundabout. Also, the type of accidents in which there was a vulnerable user such as a bicycle and motorcycle greatly increased the probability of being injured in the accidents, on the other hand, the presence of unauthorized access to the circular carriageway of the roundabout increased the severity of accidents from damage to injuries.