Abstract
Having not enough priori knowledge, it's a difficult work for a user to choose proper input parameters of a clustering algorithm. To find the best clustering result, the usual strategy is "trial-and-error" which repeats a clustering algorithm several times with different input parameters. It's well-known that clustering analysis is a time-consuming process, so repeated clustering means costing much more time. To solve this problem, this paper presents a new inheritable clustering algorithm based on K-Means, which can inherit the “good” clusters and adjust the “bad” clusters based on previous clustering results. Experiments show that the algorithm can not only get the clustering result correctly and effectively, but also avoid falling into local optimum.