基于特征点选择的聚类算法研究

A clustering algorithm based on feature point selection

针对当前数据挖掘中对数值型数据聚类方法的不足,提出了基于特征点选择的聚类算法(clustering algorithmbased on Feature Point Selection,CFPS)。CFPS算法可以克服需要输入聚类数量的缺陷,算法本身可以找到簇的最佳数量,使聚类的精度和效率得到大大提高。实验结果表明该方法对数值型数据聚类方法具有借鉴意义和深入研究的价值。

A clustering algorithm based on Feature Point Selection in Data Mining （abbreviated CFPS） is put forward in this paper. This method can overcome the disadvantage of a algorithm which requires the number of clusters for numerical incoming data. The CFPS algorithm finds the optimal number of clusters, and greatly improves the precision and efficiency of clustering. The results of experiments prove that using the algorithm for the numerical data clustering method is feasible, which is valuable for further study in more depth.