City traffic flow character analysis and origin-destination estimation based on data mining

Video detector is used widely in modern city to monitor the traffic situation. But the data from detectors is usually not fully used. In this paper, we adopt data mining method to deal with these data in order to analyze the traffic flow character and estimate OD matrix more accurately. On the basis of study data character carefully, we discuss the traffic flow feature from 3 parts: spatial, temporal and vehicle types. Then an improved method to obtain OD matrix is proposed. It costs less than traditional method and is more direct and simpler than the method which used link flow to calculate OD matrix. At last, we study a case with real data in Beijing to verify the effectiveness of this data mining method.