连续属性决策表离散化的图论方法

通过研究粗糙集与图论的关系,提出了以集合为权的加权多重完全多部图的概念,得到了加权完全多部图与连续属性决策表的映射关系,给出了断点重要性和断点效率的一种新的量化定义并得到了相关性质;提出了连续属性决策表信息系统的图论形式和连续属性决策表离散化的图论方法。编程实验结果证明,应用此方法可以确保在离散化后决策表相容的前提下得到无剩余属性值的较小的断点集合。

Graph method of discretization of decision table with continuous attributes.

Through studying the relationship between rough set and graph theory, this paper has put forward the concept of weighted complete multipartite multigraph which using set as weight, obtaining the mapping relationship between weighted complete multipartite multigraph and the decision table with continuous attributes. New quantitative definition of breakpoint importance and breakpoint effi- ciency are given and the related properties are obtained. This paper also defines the graph model of decision table information system with continuous attributes and the discretization method of decision table with continuous attributes. Experimental results show that this method can be used to obtain the smaller breakpoints set with no residual value under the premise of ensuring the decision table consis- tency after discretization.