Vehicle demand evolution analysis from the complex network perspective

Abstract

Abstract Nowadays, with the increasing population of a city, the vehicle demand especially in rush hours of weekdays increases remarkably. To dispatch taxicabs efficiently, taxicab companies such as DiDi need to analyze the dynamic vehicle demand evolution. In this work, we analyze the order data of DiDi from complex network perspective. Firstly, we propose to divide the city into many small grids. Secondly, we analyze the order data among all areas and construct a car dispatching network in which the link weight represents the number of orders from one area to another within a given time interval. Thirdly, by analyzing the network metrics such as the degree distribution, evolution of node strength and traveling time, it can be found that the results are very interesting and useful. Finally, we further discuss hub characteristic of the network, and find that more than 95% orders are served for the top 5% areas.