Options
2002
Conference Paper
Titel
Spatial subgroup discovery applied to the analysis of vegetation data
Abstract
We explore the application of Spatial Data Mining, the partially automated search for hidden patterns in georeferenced databases, to the analysis of ecological data. A version of the subgroup mining algorithm is described that searches for deviation patterns directly in a spatial database, automatically incorporating spatial information stored in a GIS into the hypothesis space of a data mining search. We discuss results obtained on a multi-relational biodiversity data set recorded in Niger. Vegetation records are georeferenced and associated with a set of environmental parameters. Further data provide information on climate, soil conditions, and location of spatial objects like rivers, streets and cities. The subgroup mining finds dependencies of a plant species on other species, on local parameters and non-local environmental parameters.