A data mining approach to support a data-driven scheduling system for air cargo terminals

Air cargo is an essential part of global supply chains when speed and reliability are the important factors. Within the last ten years, the air cargo industry is characterized by a continuous growth, which is predicted to hold on over the next years. This growth challenges the major air cargo hubs, which are the central interfaces for the handling of cargo between land and air transportation. These major terminals are often automated systems with highly integrated information systems to support the efficient handling of the air cargo in short time frames. The scheduling of the freight handling processes is a vital part of the operation processes, which can be improved when the available data and especially the information about uncertainties about previous processes is used. This paper introduces a data mining approach that automatically aggregates the historical operating data about freight handling jobs over several steps to finally train a classifier, which can be used within a scheduling system for probabilistic estimates of future job processing times.