A Discrete Route Choice Model Using Support Vector Machine an Context of Dhaka City

Abstract

The study delves to identify the route choice behavior between two alternative routes for a set of origin and destination point in Dhaka city. The Route 1 is Uttara to Motijheel through Rampura and the Route 2 is Uttara to Motijheel through Mohakhali. Perceived travel time, congestion level, trip purpose and familiarity of route are considered as important attributes for the route choice. A Support Vector Machine (SVM) approach is adopted to model the route choice behavior. The proposed model shows 88% accuracy in testing and precision of choosing Route 1 and Route 2 are 89% and 87% respectively. In total, 47.62% and 52.38% traffic are assigned in Route 1 and Route 2 respectively. Also found that, perceived travel time and familiarity of route are the most important attributes while making route choice decision.