Materialisation Forecasting: A Data Mining Perspective

  • David A. Selby
Part of the Applied Optimization book series (APOP, volume 79)


Have you ever wondered why airlines overbook planes? Or have you wondered why seat prices seem extremely variable, almost to the extent that you feel you could name the price? All of these facts come from dealing with a perishable commodity.

The airline’s dilemma is that, to survive in business, it must sell each seat, at the maximum margin, before departure. Once the door closes on a flight an unoccupied seat represents a liability. Why? Well, the cost of flying a plane is dictated by a set of fixed costs, the bulk of which is fuel and crew. Contrary to popular belief, the crew is there for safety reasons; other services, like delivering food to your seat and distributing sick bags are a matter of choice, which, in a business context, can be thought of as tailoring the product. Therefore, you cannot offload crew just to complement the number of passengers and save costs. Other costs are take-off and landing fees, ground handling, check-in, and baggage handling. Catering can be adjusted a little. Bottom-line: ensuring full planes is key to running a successful airline. But the essence of being successful in the airline business is to manage the greatest possible spread between costs and revenues. Doing so means executing effective yield management.

This chapter discusses a new approach to forecasting some of the key components in the yield management equation.


data mining multi layer perceptron neural networks yield management 


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. [1]
    Chamberlain D., Chamberlin D.D. Using the New DB2: IBM’s Object-Relational Database System. Morgan Kaufmann, 1996.Google Scholar
  2. [2]
    Christopher M.B. Neural Networks for Pattern Recognition Oxford University Press, 1995.Google Scholar
  3. [3]
    Demuth H., Beale M. Neural Network Toolbox: for Use with MAT-LAB®, User’s Guide Version 4. The MathWorks, Inc., 2001, .com/.
  4. [4]
    IBM Corporation, IBM DB2 Universal Database, Release Notes, Version 6. IBM Publication, 1999.Google Scholar
  5. [5]
    IBM Corporation, IBM Intelligent Miner Scoring Services, Version 7. IBM Publication, 1999.Google Scholar
  6. [6]
    IBM Corporation, Using IBM Intelligent Miner. IBM Publication, 1999.Google Scholar
  7. [7]
    Magnus N., Hansen L.K., Ravn O.E., Poulsen N.K. Neural Networks for Modelling and Control of Dynamic Systems: A Practitioner’s Handbook, New York: Springer-Verlag, 2000.Google Scholar
  8. [8]
    Paul Y., Haykin S. Regularized Radial-Basis Function Networks: Theory and Applications. New York: Wiley, 2001.Google Scholar
  9. [9]
    Quarendon P. The DMSS Data Mining Toolkit IBM Internal Publication, 1990.Google Scholar
  10. [10]
    Quarendon P., Todd S. The Winsom Reference Manual IBM Scientific Centre Report, 1987.Google Scholar
  11. [11]
    Selby D. Using Data mining to predict materialization for individuals. IBM RSW9–2000–0052 US1 Patent application, 2000.Google Scholar
  12. [12]
    Selby D. Using Data mining to predict materialization for groups. IBM RSW9–2000–0054 US1 Patent application, 2000.Google Scholar
  13. [13]
    Selby D. Using Data mining to assign scores for Airline flight wait list promotion. IBM RSW9–2000–0047 US1 Patent application, 2000.Google Scholar
  14. [14]
    Thomas V., Selby D. UDB Miner Reference Manual IBM Internal Publication, 1998.Google Scholar

Copyright information

© Springer Science+Business Media Dordrecht 2003

Authors and Affiliations

  • David A. Selby
    • 1
    • 2
  1. 1.Business Consulting Services, CRM SolutionsIBMUSA
  2. 2.IBM UK, North RegionWinchester, HampshireUK

Personalised recommendations