Analysis of Factors Affecting Traffic Conflict Severity at Unsignalized Intersections Using Multinomial Logistic Regression

Intersections are points where multiple traffic streams converge and, if drivers fail to follow regulations, they become hotspots for collisions. To reduce the frequency and severity of crashes at intersections, research on the factors influencing collisions is essential. Due to limitations in crash data, alternative safety assessment methods, such as traffic conflicts, are increasingly used. Most safety indicators focus on collision likelihood and temporal-spatial proximity, while assessing potential outcomes, such as injury or fatality, provides a more comprehensive measure of severity. This study evaluates factors influencing the severity of traffic conflicts at unsignalized urban intersections. Video footage from selected intersections in Tabriz, Iran, was analyzed using T-Analyst to extract variables including conflict indicators, vehicle speed, time of conflict, vehicle composition, and conflict type. A total of ۶۰۶ conflicts were identified, and multinomial logistic regression was applied to determine factors affecting conflict severity. The results confirmed a significant association of PET, ΔV۰, and vehicle mass ratio with conflict severity, with perpendicular conflicts identified as the most severe. These findings offer a practical basis for intersection safety assessment and can support traffic management authorities in implementing preventive interventions to reduce the risk and severity of crashes.

کلیدواژه ها:

Traffic Safety ، Surrogate Safety Measures ، Conflict Severity ، Unsignalized Intersections ، Multinomial Logistic Regression