A semi-automatic method to build territorial partitions
23 August 2000
Didier Josselin and Jerome Bolot
The goal of this article is to assess the suitability of a semi-automatic method to build spatial partitions. Processed data are geographical intercommunal flows, especially agricultural flows. Geographical entities are french communes to compose spatial aggregates. Spatial clustering implements two complementary aggregation methods. During a first stage, an automatic algorithm (hierarchical clustering or genetical algorithms associated to a head function) follows a defined heuristic to generate a territorial partition. Then, an exploratory method allows user to analyse geographical flows, to evaluate partition and its spatial aggregates, and to modify them interactively, if necessary. These approaches are finaly compared according to different points of view : the employed method, the built partition and the user. Consecutive partitions and aggregation methods may be used by french statistical offices, such as INSEE and SCEES, whose we keep in touch through national research programs (PSIG and INRA-DADP).
Key words : geographical intercommunal flows, spatial aggregation, genetic algorithms, hierarchical clustering, cutting out algorithms, territorial partitions, quality of geographical information, territorial pertinence, Exploratory Spatial Data Analysis, interactive GIS.
Reference: Proceedings of the Fifth International Conference on GeoComputation, University of Greenwich's School of Earth and Environmental Sciences, Kent, UK, 23 - 25 August 2000. Papers published on CD-ROM. Produced by: R.J.Abrahart and B.H.Carlisle. Publisher: "GeoComputation CD-ROM". ISBN 0-9533477-2-9