Schedule Match and Demand Projection for Future Flows

摘要

This disclosure discusses a method for distributing forecast demand to future flows based on schedule quality.The core of the network revenue management system is a forecaster function that computes an estimate of the demand associated with each future flow, i.e.Each Origin/Destination/Travel Purpose/Point of Sale/departure time.Currently this is done in two steps: the first one computes the average demand for historical "virtual flows" i.e.groups of flows otherwise identical but with sometimes different departure/arrival time due schedule changes.Secondly, demand from the virtual flows is mapped to future flows, and it is not unusual that future flows end up with no demand/too little demand/too much demand because of different errors in this process.Basically there is risk of mismatching historical virtual flows to the future flows.Also, the demand estimates of the historical flows may be wrong to begin with because of failures arising from the historical virtual flow creation algorithm.