The use of multivariate analysis techniques for exploratory spatial study of rural geospatial data has being useful in territorial characterization and typology. Connectionist approach is a very efficient computational method for these tasks, mainly the Self-Organizing Map (SOM). Through the use of SOM clustering and data visualization can be conducted simultaneously. This paper assesses the capacity of SOM algorithm in the task of exploratory multisource data analysis for local aggregation verification in the Center-South Rural Territory of Sergipe State, Brazil. Results shown that, despite the homogeneity of the region, there are significant differences among sub-groups of municipalities that compose the territory.
self-organizing maps; visualization; clustering; territory
