EXDBSCAN: An extension of DBSCAN to detect clusters in multi-density datasets

Density Density-based clustering methods are an important category of clustering methods that are able to identify areas with dense clusters of any shape and size. One of the basic and simple methods in this group is DBSCAN. This algorithm clusters dataset based on two received parameters from the user. one of the disadvantages of DBSCAN is its inability in identifing clusters with different densities in a dataset. In this paper, we propose a DBSCAN-based method to cover multi-density datasets, called EXDBSCAN. This method only get one parameter from the user and in addition of detecting clusters with different densities, can detect outlier correctly. The results of comparing final clusters of our method with two other clustering methods on some multi-density data sets shows our method´s performance in such datasets.