一种属性约简新方法--修边法

摘要

总结了粗糙集用于属性约简过程中的优缺点；为了弥补粗糙集的缺陷，构建了一套简便的属性约简新方法，即修边法；修边法类似于粗糙集，但它通过人为假设将粗糙集转化为精确集，模拟二者之间的误差率，将这种误差度与决策表的条件属性重要度等价起来；揭示了修边法中属性权重的用途和意义；通过数据实例运算对比两种方法，展现了修边法的优势；通过加入容错阈值，提高属性约简的鲁棒性。%Summarizes the advantages and disadvantages of Attributes Reduction Method in Rough Sets. In order to address the shortcomings of Rough Sets, a simple new method for attributes reduction is built, named Deburring Method. Deburring Meth?od is similar to Rough Set, but by assuming that Rough Set will be transformed into Crisp Set, the error rate between the two set is equivalent to Significance of Attributes in Decision Table. Reveals the purpose and importance of Attribute Weight in Deburr?ing Method. Comparise two methods by an instance operation, and demonstrates the advantages of Deburring Method. Improv?ing the robustness of attribute reduction by joining Tolerance Threshold.