Route choice estimation using cell phone data

Abstract

Nowadays development of cell phone network provides huge and ubiquitous data, with wide application in transportation science. One of the most important advantages of these kinds of data is enabling the process of collecting information without any active users' interference. A big data set consisting of 300,000 cell phone users' information in Shiraz are studied. This data set includes spatiotemporal information of travelers for every 5 minutes in a time span of 40 hours in two consecutive days. The spatial part of each user's information contains the position of the BTS (Base Transceiver Station) to which his cell phone is currently connected. Due to the existence of outliers, it is necessary to smooth the data initially. One of the main reasons of errors in the data set is ping pong handover, which leads to false transitions and must be eliminated. After the data preparation, stay locations are determined for each user and a trajectory for each pair of origin and destination is estimated. At this step based on network information of the city, a method to match trajectories with the network map is applied. Finally the obtained results indicate whether travelers choose the shortest path or other possible alternatives.