KNN-DBSCAN: Using k-nearest neighbor information for parameter-free density based clustering

Density based clustering is adopted in situations where clusters of arbitrary shape exist. DBSCAN is a popular density concept but suffers from the drawback of dependence on user-defined parameters like many other density based methods. In order to utilize the potential of this clustering method we propose a combination method. The information of k-nearest neighbors is used with DBSCAN to achieve a parameter-free clustering technique. The parameters are set according to information of the data as it gets accumulated in a cluster structure.