论文部分内容阅读
决策系统中连续属性的离散化,即实型属性空间向整型属性空间的映射,它是对决策表中属性约简的第一步.针对多值决策属性的决策信息系统,提出一种新的属性离散化算法.首先根据决策属性的不同,将条件属性集划分为不同的序列,对每两个序列求取候选断点,最后,综合所有的候选断点即为所求的候选断点集合;然后在基于条件属性重要度和贪心算法的基础上提出一种确定结果断点子集的新启发式算法.实例验证了本文所提出的算法能够取得较理想的连续属性离散化结果.
Discretization of continuous attributes in decision-making system, that is, the mapping of real attribute space to integer attribute space, is the first step of attribute reduction in decision table. A new decision-making information system for multi-valued decision attribute is proposed, Attribute discretization algorithm.Firstly, according to the different decision attributes, the set of condition attributes is divided into different sequences, and the candidate breakpoints are obtained for every two sequences. Finally, all the candidate breakpoints are synthesized as the candidate breakpoints Then a new heuristic algorithm is proposed based on the importance of conditional attributes and the greedy algorithm.The example verifies that the algorithm proposed in this paper can achieve the ideal discretization of continuous attributes.