基于粒化单调的不完备混合型数据增量式属性约简算法

INCREMENTAL ATTRIBUTE REDUCTION ALGORITHM FOR INCOMPLETE MIXED DATA BASED ON GRANULATION MONOTONY

增量式属性约简是一种针对动态数据集的新型属性约简方法。然而目前的增量式属性约简很少有对不完备混合型的信息系统进行研究。针对这类问题提出一种属性增加时的增量式属性约简算法。在不完备混合型信息系统下引入邻域容差关系。基于邻域容差关系的粒化单调性,提出信息系统属性增加时邻域容差条件熵的增量式更新方法,并提出了不完备混合型信息系统下的邻域容差条件熵增量式属性约简算法。实验分析表明了该算法的有效性。

Incremental attribute reduction is a new formal attribute reduction method for dynamic data sets.However,there are few studies on incomplete hybrid information systems for incremental attribute reduction.To solve these problems,an incremental attribute reduction algorithm with increasing attributes is proposed.In this paper,the neighborhood tolerance relation was introduced in incomplete hybrid information system.Based on the granulation monotony of neighborhood tolerance relation,an incremental updating method of neighborhood tolerance condition entropy was proposed when the attribute of information system was increased,and an incremental attribute reduction algorithm based on neighborhood tolerance condition entropy was proposed for incomplete hybrid information systems.The experimental results show the effectiveness of the proposed algorithm.