Characterizing and Modeling Package Dynamics in Express Shipping Service Network

Along with the increasing prosperity of market economy and the growth of online retail, express shipping service (e.g. FedEx, UPS) is playing an increasingly important role in our daily lives. A thorough understanding of the network structure and the package traffic dynamics of largescale express shipping service network (ExpressNet) is essential for performance evaluation, network optimization, and user experience enhancement. Moreover, it would also be interesting and helpful to investigate how express shipping service reflects people's daily lives. In this paper, we propose systematic work to characterize and model the traffic dynamics in a nationwide ExpressNet. We collect 16 million delivery traces over 4 months in China, and examine its characteristics from a wide range of perspective, including network structure, temporal and spatial traffic dynamics, which provide important insights into express companies to better understand the network performance. On top of that, we develop an Extended Markov Model (EMM) to capture the dynamics of package delivery process and further predict the package delay, which is a major performance metric that both customers and express companies are concerned about. Data-based evaluation shows our model can achieve 91% prediction accuracy.