属性排序的粗糙集和统计方法研究

Rough Set and Statistics Methods Based Attributes Ranking

引入实际问题需求决策表,研究属性排序的层级式计算方法,给出层级式计算的过程,并总结出层级式计算的金字塔模型。使用粗糙集属性重要度为第1层方法,位于金字塔底端,是宏观度量;基于约简的属性频次统计为第2层方法,在约简基础上结合统计方法计算属性重要程度数值度量,是宏观和微观兼有的度量;将基于约简的属性影响程度均值作为第3层方法,考量属性的微观区别并以数值度量来进行重要性比较,位于金字塔顶端。研究结果完善了粗糙集理论和应用的方法。

Practical problems demand decision table was introduced. Hierarchy type calculation method on attributes ranking was studied. The working process of the method was presented. A pyramid model for the method was summarized. Attribute significance of rough set theory was placed at the first level. It was situated at the bottom of the pyramid and was a macroscopically measurement.The second level was reduction based frequency statistics of attributes. Numerical measures of attribute importance were calculated via statistical methods on the basis of reduction theory. And it was a measurement mixed with macroscopic and microcosmic information. The third level was a method which takes the mean value of attribute impact as a measurement on the basis of reductions.This method considers the microscopic difference of attributes and compares the importance bynumerical measures. It was placed at the top of pyramid. The results of the study can improve the rough set theory and methods on application. It can also lay the foundation for applications of rough set theory.