A Voronoi-Like Model of Spatial Autocorrelation for Characterizing Spatial Patterns in Vector Data

The paper presents a computational model of spatialautocorrelation based on a Voronoi-like auxiliary structure. It shows that the Voronoi-like partition of map objects can be used to discern spatial patterns (e.g. clustered or dispersed) of geographic phenomena. In this paper, we transform the problem of characterizing the patterns for different geometry types (i.e. points, curves, and polygons) into a process of calculating spatial autocorrelation based on the auxiliary partition units. The method is shown to be successful for the designated tasks.