Discovering Transportation Mode of Tourists Using Low-Sampling-Rate Trajectory of Cellular Data

Abstract

Transportation mode detection plays an important role in transport planning and disclosing the contextual information of an individual or a group. Most existing approaches for inferring the transportation mode rely on GPS data collected from the mobile phone users, which can get more precise detection rate but in a lower scale. In this paper, we propose a framework based on data from cellular network, i.e., call detail records (CDRs), to determine the motorized transportation mode of tourists. In order to reduce the uncertainty of the low-sampling-rate trajectories getting from CDRs of tourists, an algorithm called spatial and temporal dynamic time warping (ST- DTW) is presented to conduct route matching between tourists trajectories and various routes of different transportation modes. Additionally, for a tour group, a trajectory aggregation method is used to merge the trajectories in one group so as to improve the accuracy detection. Finally, a number of interesting insights about travel behaviors of tourists in Hainan Province are given.