一种基于分类目标的启发式离散化算法

提出一种基于分类目标的启发式离散化算法, 通过该算法能够解决粗糙集理论中的连续属性离散化问题. 该算法充分考虑目标分类和属性的重要性, 在减少决策规则的同时完成了属性约简. 通过茶味觉信号的验证及与传统算法结果的比较, 验证了所给算法的有效性.

Heuristic Discretization Algorithm Based on Objective Planning

In order to solve the problem of rough sets theory in continuous attributes discretization a heuristic discretization algorighm on objective planning is proposed. With the wealthy calculation on objective planning and the importance of attributes considered, the heuristic discretization algorithm not only reduces decision rulers but also finishes attribute value reduction. The comparison between the algorithms in this paper and the other paper on teataste signals shows the algorithm in this paper produces less rulers and the short average length of rulers. The rulers in this algorithm have been decided, too. Consequently, the decision system has a high efficiency.