Density Based Algorithm for Spatiotemporal Data

Abstract

Clustering is a method to discover inherent natural structure in a set of objects involved in any phenomenon. In this study, we extended DBSCAN algorithm for spatiotemporal data by defining attribute based mass function, density function and hence modifying definition of core objects for clustering. The proposed work generalizes the concept of using an attribute to define notion of relative importance of an object to define density in the dataset. We have used a real fire dataset to validate the proposed approach. We also compare our algorithm with DBSCAN based algorithm which is extended for spatiotemporal data. The experimental results reveal that our proposed algorithm is able to identify intrinsic information based hidden clusters, which DBSCAN based algorithm is unable to identify.