摘要
本文通过0—1化和主成分分析构成的“二次转换法”对多态数据进行数量化,再采用主成分分析缩维、平方和增量法聚类、逐步判别程序检验结果,结合orford(1976)的经验方法定类,较为理想地解决了三类问题,将64个供试剖面划分为3类,其类间差异极显著,类内差异不显著。
“Two-time Transformation” was used to transform multistate data into numerical ones, principal component analysis to compress attributespace, incremental sum of square method to cluster soil profiles into hierarchy, and succesive discreminant program to examine results,in order to deal with three problems which exist in numerical classicification of soils. Three clusters, among which the differences were very significant, were gained for the 64 studied profiles. In addition, Oford's experential method of defining class was improved.
出处
《四川农业大学学报》
CSCD
1990年第3期249-254,共6页
Journal of Sichuan Agricultural University
关键词
土壤
分类
综合数值法
SOIL CLASSIFICATION
PRINCIPAL COMPONENTS ANALYSIS
CLUSTER ANALYSIS.
