Data-Driven Multi-Criteria Group Decision Making Under Heterogeneous Information

Due to the complex decision-making problems involving multiple stakeholders, the interest in group decision-making GDM approaches increases. The success of the GDM processes is directly related to the evaluations of the decision makers (DMs). When DMs have varied contributions because of their expertise, experience, consistency with others, etc., they are assigned weights to incorporate their value in the final result. In the literature, although there are many studies on criteria weights, the number of studies on DM weights is limited. In this study, a data-driven methodology is proposed to find the weights of DMs by using a machine learning (ML) method. For this, initially, an ML algorithm is designed to find the relations between the weights of the DMs and their characteristics, such as age and experience, using the weight schemes applied in previous GDM processes. Subsequently, the weights are calculated for the given problem on hand, according to the characteristics of the DMs involved. In Multi-Criteria Group Decision-Making (MCGDM) problems, DMs may provide their evaluations in different formats. In this study, to deal with such heterogeneous information cases, the cumulative belief degree (CBD) approach based on belief structure and fuzzy linguistic term is proposed. The information provided in intuitionistic fuzzy numbers, hesitant fuzzy linguistic terms, and hesitant fuzzy numbers is converted to belief degrees to find the final rankings of the alternatives. As a result, a data-driven MCGDM methodology is proposed where the weights of the DMs are calculated by using an ML algorithm and heterogeneous information is aggregated by the CBD approach. The proposed methodology is tested on the generated synthetic data.