摘要
挖掘肿瘤诊断数据库中的关联规则,能为肿瘤诊断提供有用的信息。肿瘤诊断数据库中的属性常为数量型属性,因此如何将数量型属性离散化是挖掘关联规则的难点。竞争聚集算法综合了分层聚类与划分聚类的优点,它能够有效地体现数据的实际分布情况并得到优化的聚类个数,因此能将数量型属性离散化成若干个优化的区间。
Association rules used in mining the database of tumor diagnoses can provide useful information for tumor diagnoses. Attributes in the database of tumor diagnoses are usually quantitative attributes, so quantitative attribute discretization is a problem of mining association rules. CA algorithm combines the advantages of hierarchical clustering and partition clustering techniques, it can effectively embody real distribution of data and obtain optimal number of clustering, so it can partition continuous attribute into several optimal intervals.
出处
《计算机工程》
CAS
CSCD
北大核心
2003年第12期8-9,47,共3页
Computer Engineering
基金
国家自然科学基金重点项目(69931040)
关键词
数据挖掘
竞争聚集算法
数量型属性
关联规则
肿瘤诊断
Data mining
Competitive agglomeration (CA) algorithm
Quantitative attributes
Association rules
Tumor diagnoses