摘要： 在粗糙集理论的基础上，对决策信息系统中边界区域的数据进行研究，提出一种从边界区域数据中挖掘决策规则的算法——近似序列决策规则挖掘算法。在16个UCI数据集上的测试表明，该算法在规则的准确度和平均前件长度2个指标上优于ID3算法，能简洁、高效地挖掘出决策信息系统中的全部决策规则，为挖掘未知知识提供了新的思路。针对挖掘出的全部决策规则，提出新的确定性度量和一致性度量指标，用以准确地反映决策规则的性能。
关键词:
决策信息系统,
粗糙集边界区域,
决策规则,
规则度量指标
Abstract: Extraction Algorithm of Approximate Sequence Decision Rules (EAASDR) extracting decision rule from border region of rough set is proposed. It can extract all knowledge from decision information systems. Comparison tests between EAASDR and ID3 in 16 UCI data sets show that the algorithm is prior to ID3 in the accuracy of rule set and the average condition number of rule sets. A new rule measure criterion of certainty and consistency is proposed in order to accurately reflect the performance of all decision rules extracted from decision table.
Key words:
decision information system,
border region of rough set,
decision rule,
rule measure criterion
中图分类号:
王树锋;吴耿锋;潘建国. 决策信息系统中挖掘全部决策规则的算法[J]. 计算机工程, 2008, 34(6): 22-24.
WANG Shu-feng; WU Geng-feng; PAN Jian-guo. Algorithm for Extracting All Decision Rules from Decision Information Systems[J]. Computer Engineering, 2008, 34(6): 22-24.