Multi-graded Hybrid MRDM Model for Assisting Financial Performance Evaluation Decisions: A Preliminary Work

Abstract

This study proposes a novel multiple rule-base decision-making (MRDM) model to transform the current bipolar model into a multi-graded one based on the theoretical foundation of rough set approximations. In the existing bipolar model, the decision class (DC) comprises only three classes: positive, others, and negative ones, and the induced positive or negative rules by the dominance-based rough set approach (DRSA) or variable-consistency dominance-based rough set approach (VC-DRSA) are constrained by the dominance relationship. In certain scenarios or applications, the decision attribute of a bipolar model might need to be transformed into multi-graded DCs to meet practices; examples are the commonly observed Likert 5-point scale questionnaire adopted in a marketing survey. In other words, by eliciting a decision maker’s (DM’s) preferential judgements on the preferred degree of each DC, the newly proposed model may be more flexible to reflect the DM’s preferences or knowledge on modeling an application in a more delicate manner. To reach this goal, the present study proposes a novel MRDM model with multi-graded preferential degree of each DC. Furthermore, the performance of each alternative’s score on each rule can be assessed by the crisp (i.e., binary) or fuzzy set technique (FST) and aggregated by a linear or nonlinear operator. This study provides an exemplary case by evaluating the performance of a group of financial holding companies in Taiwan by using the binary assessment and the simple additive weight (SAW) aggregator. The obtained ranking by evaluating their financial data in 2016 is consistent with their actual financial performance in 2017, which suggests the validity of the proposed model.