An Efficient Clustering Algorithm for 2D Multi-density Dataset in Large Database

Spatial clustering is an important component of spatial data mining. The requirement of detecting clusters of points arises in many applications. One of the challenges in spatial clustering is to find clusters on multi-density dataset. In this paper, a grid-based density-confidence-interval clustering algorithm for 2-dimensional multi-density dataset is proposed, called GDCIC. The proposed algorithm combines the density confidence interval with grid-based clustering, and produces accurate density estimation in local areas for local density thresholds. Local dense areas are distinguished from sparse areas or outliers according to these thresholds. Experiments based on both synthetic and real datasets verify that the algorithm is efficiently for multi-data sets and handle outliers effectively.