Analysis of Key Route Attributes for Route Choice Model Estimation Based on GPS Data

Abstract

Efficient traffic control requires an understanding of vehicle flows in the road network, where two important components are demand and route choice. The aim of this study is to use GPS data to identify key attributes explaining the route choice observed in data. The data set consisted of about 150,000 trips and was divided into a training dataset and a test dataset. The two datasets were compared and experiments show that the routes used are similar. Discrete route choice models were estimated with a route set of observed routes in the training data. The best model showed a result of 73 % correct predictions and a route choice distribution error of 19 %. Furthermore, the result indicated that simplicity of the route is more important than the travel time.