Numerical attribute reduction in decision tables based on weighted discernibility matrix

Attribute reduction is one of the key problems of rough set theory, however, classical rough set theory cannot deal directly with continuous-valued attributes. In order to effectively deal with the problem of numerical attribute reduction in decision tables using rough set theory, a numerical attribute reduction method based on weighted discernibility matrix is developed in this paper. On the basis of Rough set theory and Fuzzy set theory, this algorithm can deal with numerical attribute reduction without data discretization, as a result, loss of information during data discretization is avoided effectively. Simulation results show that the proposed approach is effective for reduction of continuous-valued attributes, and compared with the other approaches based on classical rough set theory, this method can select fewer attributes and keep or improve classification ability.