Attribute reduction algorithm of numerical information system based on weighted neighborhood entropy
In the attribute reduction of neighborhood rough set,each attribute was given the same weight and could not make better attribute selection.To solve this problem,a neighborhood conditional entropy attribute reduction algorithm with attribute weight was proposed.The weight of conditional attributes was evaluated by the correlation coefficient between conditional attributes and decision attributes.Based on the weight method,an improved neighborhood relation was proposed,which was called weighted neighborhood relation.The corresponding weighted neighborhood rough set model was also proposed.Based on the weighted neighborhood rough set model,the weighted neighborhood entropy model was further proposed,and the monotonicity of the weighted neighborhood con-ditional entropy was proved theoretically.A new attribute reduction algorithm for numerical information system was proposed by u-sing the weighted neighborhood conditional entropy as the heuristic function.The experimental results showed that the proposed at-tribute reduction algorithm had better attribute reduction performance.
numerical information systemneighborhood rough setattribute reductionattribute weightneighborhood entropy
万方数据
维普
