GREEN: A Global Optimization Scheme for Transportation Efficiency by Mining Taxi Mobility

Abstract

Taxi business, with its ubiquitous availability, route flexibility and comfortable travel experience, offers a complementary service for the public transportation system. Among the existing methods of doing taxi business, an meaningful issue is to mine efficient seeking strategies for taxi drivers, in order to improve transportation efficiency. Recent efforts have been made mainly on the individual recommendation with respect to shorter seeking time and higher seeking efficiency, whereas the global transportation efficiency will greatly be reduced once each driver only pays attention to his local optimization. Rather than the individual recommendation, in this paper we conduct research on mining the taxis mobility from large-scale taxi data, thereby proposing a novel solution, namely GREEN (short for A Global RoutEs rEcommeNdation), to improve the seeking strategies and optimize the global transportation situation. Specifically, we first investigate how the drop-off information affects seeking strategies and conduct quantitive analysis, revealing the impact of seeking efficiency, passenger density and top drivers’ experience. Moreover, to deal with the conflict between local optimization and global optimization, we dynamically adjust the weights of road segments based on the number of vacant taxis passing through each road segment. Also, to well evaluate the transportation efficiency, we define the seeking efficiency, net revenue and operation efficiency. Extensive experiments on the real-world dataset demonstrate that our scheme can work well, which not only improves the overall seeking efficiency by reducing total vacant driving time, but also increases the global operation efficiency, thereby optimizing the global transportation efficiency.