Abstract
This paper describes the applications of an enhanced neuralnetwork and genetic algorithm for bus tickets sales forecasting.The proposed approach has several significant advantagesover conventional prediction methods. The major advantage ofthe approach is that no assumptions need to be made about theunderlying function or model, since the neural network is ableto extract hidden infonnation from the historical data. Althoughneural networks represent a promising altemative forforecasting, the problem of network design remains and couldimpair widespread applications in practice. The genetic algorithmis used to evolve neural network architectures automatically,thus eliminating the pitfalls associated with human engineeringapproach.References
G. Stefancic: The Model of Advancement of the Traffic
Technology on the Bus Station. PhD thesis, Faculty of
Traffic Sciences, Zagreb, 1993. (in Croatian)
N.T. Thomopoulos:Applied Forecasting Methods. Prentice
Hall, Englewood Cliffs, N.J., 1980.
D.T. Pham, X. Liu: Neural Networks for Identification,
Prediction and Control. Springer Verlag, London, 1995.
D.E. Rumelhart, J.L. McCleUand: Parallel Distributed
Processing. Vol.l, MIT Press, Cambridge, MA, 1986.
J.H. Holland: Genetic Algorithms. Scientific American,
Vol. 267, pp. 66-72, 1992.
NeuroGeneticOptimizer, BioComp Systems, Inc., 1997.
