Revealing mobility pattern of taxi movements with its travel trajectory

Taxi movements contain much mobility information of passengers and taxi drivers in a city. However, previous researches often overlook the continuous movements of a single taxi. In this paper, we study the mobility pattern of taxi movements by analyzing the visiting sequences which record the cells (i.e., locations) being visited by taxis. At the collective level, we observe the weak power-law scaling exists in the relation between the number of visits of a cell and the corresponding number of taxis. At the individual level, we notice some unusual characteristics of continuous movements of a single taxi: (i) the number of different cells that have been visited increase with the visit number and shows a robust scaling behavior; (ii) the cell’s visiting frequency decreases with the rank of the cell and presents as a logarithm function; (iii) The distribution of the number of visits taken to revisit one cell presents as an exponential distribution. The empirical result demonstrates that the taxi movements do follow the unified pattern and show a stronger exploration tendency and weaker preferential return than other human dynamics. Finally, we utilize an agent-based model to reveal and understand the unified pattern of taxi movements. The model results indicate that the rank distribution is an intrinsic characteristic to explore the taxi destination choice and simulate the taxi movements.