基于动态矩形的聚类方法的设计与实现

Design and realization of dynamic rectangle-based clustering approach

提出了一种新的基于动态矩形的聚类方法DRCA。该方法减少了参与聚类计算的数据元素的数量,在每一次基本聚类过程中,采用数据之间空间位置比较取代复杂的聚类距离函数计算,使得算法复杂度与数据量具有近似线性时间关系。试验结果表明了DRCA的正确性和有效性。

A dynamic rectangle-based clustering approach（DRCA） was presented. The number of data that needed to be examined for clustering was reduced, and in each procedure of clustering, the position comparison between numbers was used without any distance comparison, which made DRCA had the nearly linear time complexity with the size of dataset. The experiment results show that the DRCA is correct and efficient.