A Rough Set Model for Travel Time Prediction

Abstract

Travel time prediction of urban networks was studied. Since urban travel times are stochastic and uncertain, a model for addressing urban travel time prediction by using transport information granular computing theory based on rough set was proposed. An urban route of Delft, the Netherlands, was selected as the test bed to test the proposed model. The results show that (1) feed with raw data, the model produces error of 35%; (2) with data pre-processing, the model improves performance significantly; (3) the classifications of condition and decision attributes significantly influence the accuracy. With the optimal setting of the ranges, the proposed model is able to describe traffic phenomena with physical meaning. Overall, the accuracy is acceptable.