Traffic Crash Analysis and Evaluation Using Association Rules Mining among Vulnerable Users in Urban Streets

In urban streets, severe traffic crashes mostly relate to vulnerable users such as pedestrians, and cyclists. Because of not being protected like vehicles for their occupants, the severity of such crashes is high. Now, in the world sustainable transportation (walking and bicycle) is highlighted, thus it is important to regard the safety of this group of users more. For this purpose, it is necessary to gain an insight from previous crashes of vulnerable users. Therefore it is possible to propose countermeasures to reduce the crashes and prevent future incidents. In this paper, association rules s used as a machine learning method to find the frequent patterns and rules in the crashes of pedestrians, and cyclists. Tehran (capital of Iran) as a city with high frequency of fatal crashes in the period of 2020 to 2021 is selected as the case study. The results indicated that most fatal crashes among vulnerable road users relate to pedestrians and motorcycles (43 and 42 percent). Most of those who died in pedestrian crashes were foreigners, homeless, and municipal workers (support 0.926). Most of the injuries related to the crash of vehicles with motorcycles (support 0.661). In injury crashes of motorcycles, road and environment factors did not have a great role. Nevertheless, human factor in the form of non-necessary agility and acceleration was an effective variable (support 0.661). Engineering, education, regulation and enforcement can be the most effective countermeasures to reduce the severe crashes among vulnerable road users in Tehran.