论文部分内容阅读
针对启发式约简算法难以获得最小约简的问题,研究属性之间的排斥与吸引等关联特性,给出属性重要度计算指数.在此基础上,结合属性频率方法,提出基于属性关联的启发式约简算法.该算法以最小约简为目标,采取兼顾单个属性的辨识能力以及属性之间关联的约简策略.实验结果表明,该算法比属性频率方法以及一些同类算法具有更少的属性启发次数,计算结果大部分为最小约简.
In order to solve the problem that heuristic reduction algorithm can hardly obtain the minimum reduction, this paper studies the relevancy and attraction between attributes and gives the index of attribute importance calculation. On the basis of this, combined with attribute frequency method, Reduction algorithm is proposed in this paper.The objective of this algorithm is to reduce the recognition capacity of a single attribute and to reduce the association between the two attributes.Experimental results show that the algorithm has fewer attributes than the attribute frequency method and some similar algorithms Heuristic frequency, most of the calculation results for the minimum reduction.