Predicting the development of a homogenous population in an industrializedcountry

The development of the Finnish population has been studied in order to pred ict its probable changes in the 21st century. More generally, the proposed iterative prediction procedure is useful for homogeneous populations in dev eloped countries. The Finnish population is favorable for demographic studi es because there exists accurate Finnish population data for a long histori cal period. Since several factors, for example natality, mortality, average female fertility and standard of living, have an impact on population, its modeling and prediction is an intricate matter. First, neural networks tha t are often efficient for nonlinear, complex systems were tried. However, i t was found that there were far too many input parameters and a critical sh ortage of data to train and test neural networks. Instead, a straightforwar d, iterative procedure to predict the future development of the Finnish pop ulation was created, in particular giving its probable upper and lower limi ts. (C) 1999 Elsevier Science Ltd. All rights reserved.