چكيده به لاتين
One of the topics of interest to researchers in the field of traffic safety has been investigating various factors in the severity of the accident. So far, many studies have been performed to assess the severity of injuries in traffic accidents; but, in most previous studies, the data recorded by the police at the time of the accident have been used, and the condition of each injured person has been determined only at the scene of the accident. While we know some injured people die on the way to the hospital or in the hospital or generally their status changes that their status is not considered in many studies. This can somewhat distort the results of the studies.
On the other hand, according to forensic reports, it can be concluded that despite the decrease in the number of victims, the number of injured has increased significantly in recent years, so it indicates the need to pay more attention to the injured in accidents.
Therefore, in this thesis, with the aim of investigating the consequences of each accident, the data of the accident injured were collected from the emergency organization and the hospital and linked to the accident data that police collected. Then, the injury severity model was constructed by considering the length of hospital stay as an indicator of severity. The model accuracy was 74% in all data. After constructing the decision tree model and evaluating it, the affecting parameters of the considered index (injury severity) were identified based on the Variable Importance Measure (VIM). The results show that the type of collision is the most important variable in the model made. Age category, level of education, lighting condition, and Seat belt use status were in the next important categories. Also, according to the results, the simple Bayesian Gaussian model was selected as the best among the selected models for modeling the data.
