林业中常用的一种分析方法——聚类分析

Application of Cluster Analysis on Forestry

聚类分析的基本思想是研究的样品或指标 (变量 )之间存在着的程度不同的相似性 (亲疏关系 ) ,根据一批样本的多个观测指标 ,具体找出一些能够度量样品或指标之间相似程度的统计量 ,以此为划分类型的依据 ,将一些相似程度较大的样品 (或指标 )聚合在同一类 ,而将关系疏远的聚合在不同的类 ,把不同的类型一一划出来 ,形成一个由小到大的分类系统。最后将整个分类系统绘制成一张聚类图 (或称谱系图 ) ,并由聚类图进行分类。

The basic idea of cluster analysis is that the samples or indexes studied are more or less similar, it is therefore possible to find some statistical quantities which could be used to measure the similarity between the samples or indexes. In accordance with these principles, the samples or indexes quite similar could be merged as one category, the samples scarcely related could be categorized to different categories. A category system was then obtained after all the samples or indexes are categorized. A clustering graph are worked out based on the category system obtained and category could be achieved with the information from the clustering graph.