Customized DBSCAN for Clustering Uncertain Objects

Several data management applications rely on data clustering methods which are usually designed to handle a static object as a single point in space. In recent years, clustering static objects seems to reach a stable point. Clustering uncertain objects is more challenging than clustering static objects and currently, it is actively studied in data mining clustering researches. In this paper, we study the problem of clustering uncertain objects whose locations are described by discrete probability density function (pdf). We propose to customize DBSCAN algorithm and derive formula to reduce computation cost for clustering uncertain objects. We also apply a concept of standard deviation to approximately identify uncertain model of objects. Finally, we aim to indicate how our method can be used to effectively clustering uncertain objects.