BDT-ADBSCAN: Adaptive Density-Based Spatial Clustering of Applications with Noise Based on Bayesian Decision Theory for Identifying Clusters with Multi-Densities

Density-based spatial clustering of applications with noise (DBSCAN) is a clustering algorithm based on density distributions, which is able to recognize object groups with arbitrary sizes and shapes in the constant-density datasets while failing to detect clusters with multi-densities. To overcome its drawbacks, some researchers put forward an adaptive DBSCAN algorithm by updating Eps and Minpts adaptively to find each cluster with diverse densities, but failing to discover potential clusters automatically, its limitation was still obvious. Therefore, this paper proposed a new adaptive DBSCAN algorithm based on extended Bayesian decision theory, namely BDT-ADBSCAN, to handle the problem of recognizing multi-densities and realize its automatic clustering. The experiments results show that the BDT-ADBSCAN algorithm is effective to detect datasets with varying densities automatically with higher accuracy and validity.