Cost-sensitive sequential three-way decision for information system with fuzzy decision

Sequential three-way decision (S3WD) regarded as a dynamic multi-stage decision-making model, includes three critical components: evaluation function, threshold pair, and granularity structure. In this model, the construction of granularity structure, as the basis for decision execution, is closely related to the order in which the attributes are joined, while different features may require different costs in many real-world applications. Besides, the fuzziness of label space is widespread in information systems, however, the S3WD problem of information system with fuzzy decision has received little attention. Therefore, a cost-sensitive sequential three-way decision model for the information system with fuzzy decision is proposed in this paper. Firstly, a novel granulation method is proposed based on the density neighborhood for information system with fuzzy decision, which can obtain hidden connections of different objects. On this basis, a new evaluation criterion of attribute significance is presented by considering attribute dependence and test cost simultaneously. With three different test cost distributions, a cost-sensitive sequential three-way decision model is proposed by optimizing the information granularity. Finally, experimental results indicate that the performance of this model is affected by the granulation parameter. Compared with other S3WDs, the proposed model can achieve better classification performance with lower test costs. (C) 2022 Elsevier Inc. All rights reserved.

Automated summary

This paper proposes a cost-sensitive sequential three-way decision model for information systems with fuzzy decision, which achieves better classification performance and lower test costs by optimizing information granularity.