Aggregation Bias in Traffic Flow Time Series: The Effects of Ignoring It

Current ITS infrastructure provides the ability to simultaneously capture and monitor a variety of transportation data in very fine intervals. However, in order for the data to be analyzed and integrated to the different intelligent traffic management and traveler information systems, the optimum data resolution should be selected. In many transportation studies it is suggested that data should be aggregated in order to avoid the effects of noise and spurious oscillations. In the present paper, we analyze the effect of data aggregation on the fractional dynamics of traffic volume and occupancy, as well as the information degradation between the original and the aggregated data. Results indicate that aggregation may suppress the long memory characteristics of traffic flow. Moreover, uncertainty associated with the aggregated data with respect to the original data series increases with coarser data resolutions.