Bus arrival time prediction with real-time and historic data

Provision of accurate bus arrival information is vital to passengers for reducing their anxieties and waiting times at bus stop. GPS-equipped buses can be regarded as mobile sensors probing traffic flows on road surfaces. In this paper, we present an approach that predicts bus arrival time in terms of the knowledge learned from a large number of historical bus GPS trajectories. In our approach, we build time-dependent path-section graph, where a path-section is a road segment between two adjacent bus stops, to model the properties of dynamic road networks. Then, a clustering approach is designed to estimate the distribution of travel time on each path-section in different time slots. Finally, bus arrival time is predicted based on the path-section graph and real-time GPS information. Using a real-world trajectory dataset generated by 1000 buses in a period of 2 months, a bus arrival time prediction system is built. Then we evaluate the system with extensive experiments and realistic evaluations. Experiments show that our method is close to the actual value and better than some typical algorithms.