DBSCALE: An efficient density-based clustering algorithm for data mining in large databases

Abstract

This work presents a novel clustering algorithm that incorporates neighbor searching and expansion seed selection into a density-based clustering algorithm. Data Points that have been clustered need not be input again when searching for neighborhood data points, and the algorithm redefines eight Marked Boundary Objects to add expansion seeds according to far centrifugal force, which increases coverage. Experimental results indicate that the proposed DBSCALE has a lower execution time cost than DBSCAN, mBSCAN and KIDBSCAN clustering algorithms. DBSCALE has a maximum deviation in clustering correctness rate of 0.29%, and a maximum deviation in noise data clustering rate of 0.14%.