基于灰关联分析的连续值属性约减算法

Continuous-valued attributes reduction algorithm based on gray correlation

针对目前大多数属性约减算法只能用于离散值决策表的情况,将条件属性与决策属性的关联度作为属性约减的重要性测度,同时基于条件属性间的关联度和重要度定义了条件属性的重叠性测度,据此对条件属性进行去重叠化处理,提出了一种基于灰关联分析的连续值属性约减算法CARAG,实现了对连续值属性集的约减,并在仿真实验中对算法的可行性和有效性进行了对比验证。

Since most current attributes reduction algorithm can be only used for discrete decision tables, the correlation degree between condition attributes and decision attributes was defined as the importance degree of attributes, and meanwhile the overlap degree was defined based on the correlation degree and importance degree among attributes. The condition attributes＇ importance was renewed according to the overlap degree. To achieve the reduction of continuous-valued attributes set, an attributes reduction algorithm based on gray correlation analysis was proposed. The feasibility and effectiveness of the algorithm were verified in the simulation.