摘要：DBSCAN is a popular density-based clustering algorithm in data mining. However, its principle of first come, first served to deal with the cross-border points of multiple clusters will make some cross-border points not belong to the best clustering. To solve this problem, an adjusting strategy after DBSCAN to change the positions of the points of every cluster’s convex hull is proposed in this paper. At the end, the effectiveness of proposed adjustment strategy is verified by an experiment.
