Generating Interesting Alternatives in GIS and SDSS Using Genetic
Algorithms
Dibble, Catherine and Paul J. Densham (1993) Proceedings GIS/LIS'93, Minneapolis, Minnesota, November 1993.
Abstract
Decisions often are evaluated on the quality of the process that
supported them. It is in this context that GIS and SDSS
increasingly are being used to generate alternatives to aid
decision-makers in their deliberations. Unfortunately, GIS and
SDSS typically lack formal mechanisms to help decision-makers explore
the solution space of their problem and thereby challenge their
assumptions about the number and range of options available. We
describe the use of a genetic algorithm to generate a range of feasible
alternatives to location selection problems. The ability of
genetic algorithms to search a solution space and selectively focus on
promising combinations of criteria makes them ideally suited to such
complex spatial decision problems. We also describe the
implementation of this algorithm in a microcomputer-based SDSS and
present representative results for several location selection problems.
Finally, we discuss the inherent parallelism of this algorithm
and strategies for its decomposition that will enable it to exploit the
efficiency gains of parallel processing computers.
27 June 2003