Abstract
An application of a parallel genetic programming approach to discovering new spatial interaction models is described. It is noted that geographic information systems have resulted in the creation of extremely data-rich environments but it is proving very difficult to exploit this situation because of the lack of suitable models and the difficulties of model creation by more traditional hypothetico-deductive routes. The author describes how to develop new models via a machine-based inductive approach. A Cray T3D parallel supercomputer with 512 processors is used to investigate the potential of this approach to building computer models. Results are described which allude to the potential power of the method for applications where there is a considerable amount of data but no suitable existing models and no good theoretical framework on which to base their development.
Get full access to this article
View all access options for this article.