一种基于t相似分类的连续值域决策表的决策算法

• 作 者: 刘文军 谷云东 李洪兴
• 机 构: 北京师范大学数学科学学院,
• 出 处: 模式识别与人工智能
• 发表日期: 2005年06期
• 关键词: 连续值域决策表 t相似度 t相似类 相似划分算法 广义决策集 规则获取

首先给出t相似度的定义,并引入t相似类的概念。随后,借助研究对象的t相似类定义了相似划分算法和广义决策集,并研究它们的性质,给出基于相似划分的类对应规则提取方法。接着,给出类对应规则中各条件类的区间表达,得到了面向连续值域决策表的规则提取算法。最后,结合实例说明了决策规则提取算法的实现过程。

First, we give the definition of t similarity and introduce the concept of t similar class. Subsequently, similar segmentation algorithms and generalized decision-making sets are defined by means of the t-similarity classes of the research object, and their properties are studied. A class-based correspondence rule extraction method based on similarity division is given. Then, the interval expression of each class in the class correspondence rule is given, and the rule extraction algorithm for the continuous range decision table is obtained. Finally, an example is given to illustrate the implementation of decision rules extraction algorithm.