基于邻域关系提出一种综合考虑正域和边界数据的属性约简方法.该方法利用邻域关系对数据进行离散化处理,通过定义基于邻域的正域属性重要度、边界属性重要度和邻域综合属性重要度概念,设计一种新的启发式属性简约算法.该算法从空约简集出发,利用邻域属性重要度启发式搜索属性空间以扩展约简属性集,理论分析和实验表明该算法有效可行.
Based on neighborhood relations, a new attribute reduction method based on the comprehensive consideration of domain and boundary quality is introduced. The method uses the neighborhood relationship to discrete data by defining the importance of positive and boundary region and their fusion, and results in the design of the new heuristic attribute reduction algorithm. Theoretical analysis and experimental result show that the algorithm is effective and feasible.