Sayed Amin Nemati; Reza Amin; Ali Khodaii
Abstract
The present study focuses on the binary logit model for predicting the probability of driver fatalities in road accidents. The study utilizes data from road traffic accidents in Canada. These data were collected by the highway police in 2019 and recorded in the National Collison Database (NCBD). The ...
Read More
The present study focuses on the binary logit model for predicting the probability of driver fatalities in road accidents. The study utilizes data from road traffic accidents in Canada. These data were collected by the highway police in 2019 and recorded in the National Collison Database (NCBD). The dependent variable in this model is the severity of accidents, which is a binary variable representing driver fatalities and injuries. The independent variables include vehicle types, vehicle age, days of the week, time intervals, same-direction and opposite-direction collisions, intersections, weather conditions, driver age, and gender. By analyzing the data and estimating the parameters, the model can predict up to 41% of the variations in the dependent variable. In the model validation stage, the data were divided into two parts, with 70% used for modeling and 30% for validation. McFadden's pseudo were used to evaluate the model's performance. The models were constructed using SPSS and Nlogit6.0 software. The results demonstrate that the model fits well with the data and has the capability to predict changes in accident severity. Consequently, the study indicates that variables such as road dryness, midnight time interval, and vehicle age contribute to an increase in driver fatalities, while variables such as light duty vehicles, school buses, and same-direction collisions contribute to a reduction in fatalities.
