Rough-Set Based Hotspot Detection in Spatial Data

Abstract

A special type of cluster is called hotspots in the sense that objects in the hotspot are more active as compared to all others (appearance, density, etc.). The object in a general cluster has a similarity which is less than the object in the hotspot. In spatial data mining hotspots detection is a process of identifying the region where events are more likely to happen than the others. Hotspot analysis is mainly used in the analysis of health and crime data. In this paper, the health care data set is used to find the Hotspot of the health condition in India. The clustering algorithm is used to find the hotspot. Two clustering algorithm K-medoid and Rough K-medoid are implemented to find the cluster. K-medoid is used to find the spatial cluster, Rough K-medoid finds the cluster by removing boundary points and in this way find cluster which is denser. Granules are created on the clusters created using K-medoid and Rough K-medoid and point lying in each granule is counted. Granule containing points above a particular threshold is considered as a potential hotspot. To find the footprint of the hotspot convex hull is created on each detected hotspot. Also in this paper hotspot and footprint is defined mathematically.