Abstract
Real-time traffic prediction is a fundamental capability of reducing traffic congestion and improving traffic mobility. Recently, the traffic information in the urban area of the Taipei city is available from Institute of Transportation, Ministry of Transportation and Communications, Taiwan. Given this information, it is possible to analyze and extract some traffic patterns. In this paper, we use these patterns to design a semi-parametric prediction model which provides an efficient mean to estimate accurately the future traffic conditions. Furthermore, we propose a novel vehicle routing algorithm which can plan routes with less delay for the drivers. The vehicle routing algorithm is composed of our proposed prediction model and the existing shortest path algorithm. Finally, in the performance evaluation, we show the capability of our methodology to predict future traffic conditions accurately and to enable the drivers to arrive the destinations within less time.