Ahmad Fariullah Omid; Reza Amin; Ali Khodaii
Abstract
Road accidents pose a persistent challenge to transportation systems, costing approximately 3% of the gross national product of countries annually. In this research, a multinomial logit model (MNL) was employed to assess variables associated with accidents involving private cars, taxis, buses, motorcycles, ...
Read More
Road accidents pose a persistent challenge to transportation systems, costing approximately 3% of the gross national product of countries annually. In this research, a multinomial logit model (MNL) was employed to assess variables associated with accidents involving private cars, taxis, buses, motorcycles, and bicycles. Private car accidents were considered as the reference category, and the multinomial logit model was used to create a more accurate and data-driven model. The results indicated that the use of bicycles and motorcycles had a higher likelihood of accidents compared to private cars, while buses and taxis had a lower likelihood. The data used in this study were collected by the Traffic Transport Organization of Manchester, England, from the years 2010 to 2021. The statistical software SPSS was utilized for model construction, and accidents were evaluated based on the type of vehicle. Out of the total of 21 variables used, 13 variables were found to have a significant impact in the model. The Log likelihood (LL) was employed to evaluate the models, resulting in a value of 58998.662 for the first model and 80427.09 for the second model. Considering the higher Log likelihood value in the second model, it was selected as the superior model in this research. The findings revealed that variables such as deviation from the route, personal trips, and driver gender contributed the most to accidents.