Attribute Reduction of Incomplete Information Systems: An Intuitionistic Fuzzy Rough Set Approach

Nowadays, fast expansion of data processing tools leads to increase in databases in terms of objects as well as attributes in different fields like image processing, pattern recognition and risk prediction in management. Attribute reduction is a process of selecting those attributes that are mutually sufficient and individually necessary for retaining basic property of the given information system. In this paper, we introduce a novel approach for attribute reduction of an incomplete information system based on intuitionistic fuzzy rough set theory. We define an intuitionistic fuzzy tolerance relation between two objects and calculate rough approximations of an incomplete information space by using tolerance classes of each object. The degree of dependency method is used for calculating reduct set of an incomplete information system in order to handle noise and irrelevant data. An algorithm is presented for better understanding of the proposed approach and is applied to an incomplete information system. Finally, we compare proposed approach with an existing approach for attribute reduction of an incomplete information system through an example.