Traffic estimation on full graph of transport network using GPS data of bus movements

Fast and accurate assessment of the road network loading on the city level is necessary for a number of applications, including operational control in cargo logistics, passenger transportation and work of emergency services. One of the prominent ways to achieve it is to use transport vehicles with geolocation equipment on board to probe traffic conditions. The issue is that the distribution of devices among the vehicles operating in the network is at times sparse and uneven. This creates the necessity to reconstruct actual traffic conditions from incomplete data. In this paper, we address the issue with suggesting the methodology capable of reconstructing traffic conditions for the whole road network from GPS tracks generated by public transportation vehicles. The method has been tested for the dataset collected for the road network of Saint Petersburg. The proposed approach allows to calculate time of a ride between two pairs of addresses located not more than 15 kilometers (in a straight line) apart with the Mean Absolute Percentage Error of just over 20 percent.