Dominance-based decision rule induction for multicriteria ranking

We consider the issue on decision rules induction for multicriteria ranking. Multiple Criteria Decision Analysis (MCDA) aims at giving people the knowledge of recommendation concerning a finite set of objects evaluated with multiple preference-ordered attributes (known as criteria). Dominance-based Rough Set Approach (DRSA) is a powerful tool for MCDA via assigning objects to several predefined and preference-ordered decision classes. Most of previous strategies are to induce a minimal set of "if...then.'' rules. In this paper, we provide strategies to induce a new rule set as the substitution for the classical minimal rule set. The main contributions include: (1) providing methods to induce certain rules in two situations respectively: multi-criteria and mix-attributes; (2) providing the concept of believe factor and its three measuring degrees for exploring valuable uncertain information within rough boundary regions; (3) providing the properties of believe factor with explanations from the viewpoint of class-based rough model; (4) proposing an extended Net Flow Score method in consideration of both partial and total orders in multicriteria ranking, via our proposed decision rules. A numerical example is used for illustration of overall problem-solving procedures and for a comparison with the existing representative proposals.