Predicting show rates in air cargo transport

Overbooking is an important tool for revenue optimization in airline industry both, for passenger and cargo transportation. While the former is 'binary and one-dimensional' as the passengers either show up or not, the latter is more difficult. In particular, a commodity might show up for transport but both, its actual weight and volume, might differ significantly from the values specified in the booking. A reliable prediction of the show rates is therefore instrumental for any reasonable revenue optimization in air cargo industry. The present paper presents a new mathematical optimization model for predictive analytics. The exposition focusses, on the one hand, on the theoretical background of our approach which combines statistics, diagrams, clustering and data-transformations. On the other hand, we report on the successful application on (near) real world data from air cargo industry.