The Research On Cost-sensitive Three-way Model For Hybrid Feature Data And Fuzzy Decision Data

In the era of big data,uncertain and fuzzy data widely exists in all areas of real life,and the three-way decision is one of the effective tools for dealing with such uncertain decisions.From the perspective of granular computing,it divides the universe into three two-pair disjoint universe subspaces(positive region,boundary region,and negative region),and adopts different corresponding strategies(acceptance,non-commitment,and rejection)for objects in the three regions.The idea of “Trisection-and-acting” can effectively solve the problems of uncertainty and incompleteness of information.With the rapid growth of the number of data,the forms of data show diversification and complexity,while the research objects of the existing three-way decision models are mostly decision-making systems with single data.There are relatively few studies on hybrid data and fuzzy decision information systems.Therefore,in order to expand the three-way decision model and make this model more universal,this paper conducts research on the cost-sensitive three-way decision model based on rough sets theory and granular computing.The main contributions of this paper are as follows:1.In view of the problem of boundary region classification for hybrid decision system.First,the core attribute set of is computed by positive region reduction in the hybrid neighborhood decision system.On this basis,the hybrid neighborhood class is calculated and the objects are divided into the positive region,boundary region and negative region of each decision class through three-way decision rules,so as to a three-way decision model for hybrid data is proposed.Then,the classification method of three-way decision boundary region based on misclassification cost is proposed by combining cost-sensitive learning.A calculation method of misclassification cost is constructed to further divide the objects in the boundary region.And examples are given to exhibit the process of the three-way decision boundary region classification.Finally,the effectiveness of this method are verified by experiments results,and provides a method that can be used for reference for the classification of three-way decision boundary region.2.Aiming at the granularity optimization of sequential three-way decision in fuzzy decision information system.The fuzzy decision information system is granularized based on the density neighborhood to obtain different decision classes.On this basis,a new measurement criterion of attribute significance is constructed by jointly considering attribute dependence and test cost.Then,based on three different test cost distribution functions,above metric criterion is used to construct a sequence of attribute subsets in the granularity structure,a cost-sensitive sequential three-way decision model is proposed based on optimizing the information granularity.And examples are given to demonstrate the optimization process of sequential three-way decisions in fuzzy decision information systems in detail.Finally,the feasibility of this model are also illustrated,and the scope of applicability of the sequential three-way decision model is expanded.