以平均距离为基础的两种多元等级聚合新策略

TWO POLYTHETIC-AGGLOMERATIVE CLUSTERING STRATEGIES

本文基于扩展的Ｌａｎｃｅ和Ｗｉｌｌｉａｍｓ的多元聚合模型之上，提出两种新的聚合策略：新组内平均距离法和加权平均距离法．这两个方法既考虑了组间距离，同时也考虑组内的同质性，聚类结果更好．本文用山西中条山荆条灌丛的数据作为这两种方法的应用实例，结果较好地描述了群落类型间的关系，生态意义明确．这两个方法分辨力较强，适合群落低级单位的分类，宜于群落生态关系的研究．

Based on extended Lance and Williams' model, two polythetic agglomerative dustering strategies, i.e. the Average distance of new cluster and the Weighted average distance,are introduced in this paper. As an example, vegetation data of shrub, Vitex negundo var.heterophylla, collected from Zhongtiao mountains in Shanxi Province were analysed with these two new methods. The Results illustrate that the two methods are effective and successful in describing the relationships between vegetation clusters. They have some advantages compared with fuzzy equivalence clustering and group averaging, and are partically suitable for vegetation classfication at association level.