Handling spatial-correlated attribute values in a rough set

Abstract

Rough set theory has been widely used in spatial analysis. However these applications take little account of the spatial characteristics of spatial data, especially spatial dependencies and correlations. This paper proposes a new method to consider spatially correlated information in rough sets theory. This method divides the attributes of geographical objects into two categories, namely spatial correlated attributes and non-spatial correlated attributes. These two types of attributes are handled separately and the results from both types of attributes are then combined to generate the decision rule. An example is given to illustrate how the new method handles spatially correlated information in rough set theory. © 2009 Springer Berlin Heidelberg.