Improved varied density based spatial clustering algorithm with noise

Abstract

VDBSCAN is very famous Density based clustering algorithm. Handling highly dense data point is a challenging task in clustering. VDBSCAN algorithm handles widely varied density data points well and also over comes the problem of noise and outlier. But this algorithm is depends on the input parameters Eps and Minpts. The careful selection of these input parameters plays an important role in proper clustering. We propose automatic parameter selection in VDBSCAN for perfect clustering. Synthetic data with 2-dimention is used for the experiment. The result shows that, the proposed work enhances VDBSCAN algorithm.

Study of Density based Algorithms

“…DBCLASD, OPTICS DBSCAN are density based clustering algorithms. [4] Different algorithms in density based clustering like UDBSCAN [8], GDBSCAN, P-DBSCAN and current algorithms OPTICS are discussed and studied [1]. Three different algorithms of same categories i.e.…”