Attribute Reduction Of Decision Rough Set Based On Local Idea

The rough set theory was first proposed by Polish mathematician Pawlak in 1982,which is mainly used to deal with the uncertainty.However,practically,the data often have the characteristics of complex structure and high dimension.Using data directly in the application is often impractical,it follows that the number of the attributes of data sets should be reduced.Attribute reduction is an indispensable and important research content in rough set theory,and it is also one of the main contents in this paper.Some parts of decision rough set are explored in this paper.Firstly,a new heuristic algorithm for attribute reduction is proposed in decision rough set.Secondly,the multiple cost decision-making rough set model is studied.Finally,the concepts of pessimistic decision rules and optimistic decision rules are introduced.The related experiments are conducted to describe the similarities and differences of the two models.The innovation of this paper has the following two aspects:(1)In the set of conditional attributes,the contribution of each property to the information system is different,and even some attributes prove redundant.Attribute reduction can accelerate the acquisition of rules and achieve the goal of simplifying the system,so it is of great significance to study attribute reduction.Attribute reduction requires that the attribute set classification ability remains unchanged,and some redundant attributes are deleted.This paper introduces two algorithms that are most commonly used in attribute reduction,based on the attribute importance heuristic algorithm and the attribute reduction algorithm based on discernibility matrix,and combining the two algorithms,an improved heuristic reduction algorithm is proposed,in which the result of attribute reduction can be effectively achieved better.(2)In the traditional decision rough set,a pair of threshold values are generated by the cost matrix,and the concept of the upper and lower approximation of the decision rough set is given by using the threshold value.Because the cost is not equal in the different environment,in the traditional decision rough concentration cost matrix has only one,does not consider the price change,therefore this article has discussed the multiple cost situation.This paper introduces the definition of pessimistic multiple cost decision rough set and optimistic multiple cost decision rough set,in which the threshold value is generated by multiple cost matrices,the upper and lower approximations are defined by the threshold value,and the attribute reduction problem in the rough set of multiple cost decision is discussed.In the attribute reduction of decision rough set,a heuristic local attribute reduction method is proposed from the individual decision class.