Pattern Extraction and Temporal Evolution of Short-Term Traffic Volume

Traffic volume patterns and their temporal evolution are one of the most important issues for traffic prediction and traffic condition estimation.However,little work has been conducted on identifying and associating traffic pattern occurrence with prevailing traffic conditions.In order to extract the patterns hidden in traffic volume fluctuation as well as their temporal evolution,we propose a three-layer strategy that first segments the volume into subsequences.Then,we use the recurrence qualification analysis to determine the statistical characteristics of the subsequences and the k-means clustering is used to get the hidden traffic patterns finally.A case study using three typical weekly traffic volume data acquired from a freeway in Minnesota of USA shows that the proposed method is useful for identification of the traffic pattern,and traffic prediction as well.