Prediction of Suburban Road Traffic Crashes in Guilan using Logit Model

In recent years with the rapid growth of population and increasing the number of applicants for the purchase of vehicles, traffic crashes have increased. This problem will worsen when the development and repair of roads occur at a much slower rate. High rates of road accident victims are recognized as one of the greatest safety issues in developing countries. Therefore, identifying the factors affecting crashes, which leads to a reduction in the frequency and severity of crashes, is clearly understandable. The purpose of this paper is to statistically analyze vehicle and pedestrians crashes on Chaboksar-Lahijan road and make aprediction model. Therefore, in this research, by collecting the data of 2014, 2015 and 2016 of Chaboksar-Lahijan road crashes, the frequency of effective variables on crashes has been investigated and the probability of occurrence of each category of crashes is calculated using the logit model .Accordingly after encoding independent and dependent variables and inserting them into SPSS software, the logit model was built separately for vehicle crashes and pedestrians. After modeling it was found that the backward method has a relatively high prediction power in all cases. Also, the most common types of collisions in fatal and injury crashes in Chaboksar-Lahijan road is the rear-end and front-to-side collision which indicates the need to pay attention to set the laws to pay attention to the front and observe the right of way.