Understanding road performance using online traffic condition data

Traffic conditions predominantly follow a number of recurrent spatiotemporal patterns. An understanding of these recurrent patterns is of great importance to judge whether a road section is likely to be congested, which we called 'road performance'. Traffic condition data, which are freely available online, provide valuable, full coverage, real-time information on traffic conditions. Here, we propose a method for evaluating road performance using the sample points from traffic condition data acquired from Baidu map service for free. By treating human mobility as tidal currents and daily routines as planetary gravities, we utilize a tidal harmonic analysis to establish a simulation model with MATLAB. This model can deal with traffic simulations for a long experimental period in order to obtain a usual road performance. Data from two spatial scales, covering Yunnan Province in China and its capital city, Kunming, were used to evaluate the accuracy of this model. We used 31 days of the dataset to understand the road performance at each scale via three indices: the expectation, variance, and daily patterns of traffic conditions. Our experimental results demonstrate that the tidal current approach to model traffic pattern changes in small and large spatial scales is feasible and promising.