Efficient and Accurate Traffic Flow Prediction via Incremental Tensor Completion

Timely and accurate prediction of traffic flow plays an important role in improving living quality of the public, which greatly influences the policies and regulations to be enforced and abided by. In this paper, we propose to model urban highway traffic data with an incremental tensor structure to exploit all available feature aspects. It is conceived on the solid basis of dynamic tensor model for traffic prediction, and a fast low-rank tensor completion algorithm, equipped with gravitational search algorithm, is harnessed to optimize the parameters. The proposed method excavates the inner law of traffic flow data by taking account of multi-mode features, such as daily and weekly periodicity, spatial information, and temporal variations, and so on. Empirically, multi-view experiments demonstrate the superiority of Trapit, and indicate that the proposed method is potentially applicable in large and dynamic highway networks.