GEO-SOM and ITS integration with geographic information systems

Clustering geographically referenced data is an important issue in Geographic Information Science. Although the standard SOM can be used in many of these problems, it is useful to have a clustering tool that takes into account the special importance that geographic location has in these problems. In this paper, such a tool, named GEO-SOM, is presented. The differences between the training and mapping algorithms of the standard SOM and GEO-SOM are pointed out, and some simple examples of applications are given. Another important issue in the analysis of geo-referenced data is visualization of results, and integration with well established Geographic Information Systems (GIS). It is shown that GEO-SOM can easily be integrated in such systems, and examples of relevant visualization tools are presented. The fundamental assumption of the GEO-SOM is that some variables (in this case geographical coordinates) are more important, in the sense that they condition any subsequent clustering.