Predicting the number of freeway Discretionary lane changes using counting regression models

In general, freeways, due to geometric conditions, the speed and number of overpass lines, pass a large percentage of traffic flow volume. Lane change phenomenon can decreases their functional capacity and as a result, the travel time increased traffic flow and reduces speed and safety. Line change is divided into two types of optional and mandatory line change. Mandatory lane change (MLC) occurs when a driver must change lane to follow a specified path and Discretionary lane change (DLC) occurs when a driver changes to a lane perceived to offer better traffic conditions, he attempts to achieve desired speed, avoid following trucks, avoid merging traffic, etc. The aim of this research is predicting the number of the Discretionary lane change in freeways with count regression models.

The information used in this study has been processed by the data of an interstate freeway in America. Then, using this data, two types of linear regression models, poisson regression has been paid and then, for a good evaluation of the models are presented their observation and estimation charts. One of the most important usage of this model is its use in macroscopic simulation software and use in service-level design models.

The virtual variables of the number of crossing line, the number of vehicles on the current crossing line and then the average speed of the current crossing line and the target crossing line have the most impact on the estimation of this model.

In the comparison between the two linear regression and count regression models, it was observed that The count regression model describes this phenomenon better because of the proximity of the characteristics to the line change feature. It is also suggested that in future research should also be considered the mandatory line change and the differences between the two models.