属性重要程度的测量方法及比较研究  

A Study on the Measurement and Comparison of the Importance of Attribute

作　　者：刘云霞[1] 朱建平[1] 

机构地区：[1]厦门大学经济学院计划统计系,福建厦门361005

出　　处：《经济经纬》2005年第6期58-61,共4页Economic Survey

基　　金：国家社会科学基金(03BTJ014)

摘　　要：属性重要程度的测量对于数据库的压缩和综合评价都是非常重要的。属性重要程度的测量根据数据库形式的不同有不同方法：粗糙集理论、信息增益技术；在单向有序列联资料下的以秩效应为标准的测量方法和改进的秩和法。改进的秩和法不仅分析了属性的重要程度，而且其结果还可以进行多方面的统计分析。具有较高的实用价值。The measurement of the importance degree of attribute is very important to the compression and comprehensive appraisal of database. The methods of the measurement of the importance degree of attribute are different according to the different forms of database ： rough sets, information gain technology, the measurement method of taking order effect as standard in the case of unidirectional orderly contingency data and the improved rank sum method. The improved rank sum method not only analyzes the importance degree of attribute, but also its result can be analyzed in a statistical way in many aspects, so it is of great practical value.

关 键 词：粗糙集 列联表 秩效应 改进的秩和法 

分 类 号：F224.0[经济管理—国民经济]