摘要
距离系数是传统聚类分析方法中常用的相似程度统计量。在该系数的基础上,提出了一种既能考虑到样本之间的值贴近程度,又能考虑到样本之间的形贴近程度的改进统计量——贴近度,并将其应用到实际样本数据的聚类分析中。实例表明该统计量应用于传统聚类分析是有效的。
A new statistic named similar degree is formulated by combining two kinds of statistic-distance factor and similarity factor. Because of including two parameters, value parameter and shape parameter, the new statistic could describe not only the degree of value similarity, but also the degree of shape similarity between two swatches. Based on this new statistic, the actual data is analyzed. The expermental result shows the new statistic is effective.
出处
《计算机工程与设计》
CSCD
北大核心
2006年第8期1321-1323,1368,共4页
Computer Engineering and Design
基金
国家863高技术研究发展计划基金项目(2002AA4Z3240)
关键词
聚类分析
统计量
贴近度
模糊关系矩阵
谱系图
cluster analysis
statistic
similar degree
fuzzy relation matrix
pedigree chart
