A novel extension method of vpfrs mode for attribute reduction problem in numerical decision tables

Abstract

Attribute reduction is an essential application of the Rough Set (RS) theory that has been receiving the attention of many researchers. Up to now, attribute reduction methods to improve classification accuracy on noisy datasets following the IFRS approach still have many limitations in terms of computation time. In this paper, we use the variable precision method on approximate operators of the FRS model to expand measures to effectively evaluate attributes on noisy datasets. The main contributions of this paper include: 1) proposing new approximation operations to extend VPFRS to VPOFRS, 2) proposing some measures based on VPOFRS to evaluate the consistency degree of the decision table and the dependence degree of the attribut, 3) proposing an attribute reduction algorithm VPOFRS_AR. Experimental results on noisy datasets from UCI show that the proposed method not only improves the noise for the reduct but the algorithm's execution time is faster than other algorithms.