A Clustering Algorithm for Discovering Varied Density Clusters

---------------------------------------------------------------------***--------------------------------------------------------------------Abstract Spatial data clustering is one of the important data mining techniques for extracting knowledge from large amount of spatial data collected in various applications, such as remote sensing, GIS, computer cartography, environmental assessment and planning, etc. many useful spatial data clustering algorithms have been proposed. DBSCAN is the most popular density clustering algorithm, which does not limit itself to shapes of clusters and handles the noise effectively. However, DBSCAN has a trouble in finding out all clusters from datasets with varied densities, because it depends on a globular value for its parameter Eps. This paper presents enhanced DBSCAN which clusters spatial databases that contain clusters of varying densities effectively. The idea is to allow varied values for the Eps parameter according to the local density of the starting point in each cluster. The clustering process starts from the highest local density point towards the lowest local density one. And the value of Eps varies according to the local density of the initial point in current cluster. For each value of Eps, DBSCAN is adopted to make sure that all density reachable points with respect to current Eps are clustered. At the next process, the clustered points are ignored, to avoid merging among denser clusters with sparser ones. Varied synthetic datasets in 2-dimension are used to evaluate the efficiency of the proposed enhanced DBSCAN algorithm.