BIRCH混合属性数据聚类方法

Heterogeneous data clustering algorithm of BIRCH

数据聚类是数据挖掘中的重要研究内容。现实世界中的数据往往同时具有连续属性和离散属性,但现有大多数算法局限于仅处理其中一种属性,而对另一种采取简单舍弃的办法丢失聚类信息和降低聚类质量。一些能处理混合属性的算法又往往处理的属性过多,导致计算量的大增。提出了一种基于BIRCH算法的混合属性数据的聚类算法;在UCI数据集上的实验表明,文中提出的算法具有较好的性能。

Data clustering is an important issue in data mining.Many real-world data have both continuous attributes and categorical attributes,which are usually called heterogeneous attributes.However,most of the existing mining algorithms can manipulate only continuous attributes or categorical attributes.Simply omitting categorical or continuous attributes may lose important information about the data and decrease the mining quality.Some other algorithms which can manipulate continuous attributes and cate- gorical attributes have low efficiency,because of a lot of attributes.This paper proposes a novel approach for clustering data with heterogeneous features based on BIRCH.Experimental results on public data sets show that the proposed algorithm is robust.