Integrating external representations and internal patterns into dynamic multiple-criteria decision making

The dynamic multiple-criteria decision-making (MCDM) approach has drawn immense attention in the decision analysis domain over recent years. An essential problem of this dynamic process is how to retrieve the utmost from historical data. Although most previous dynamic MCDM studies process historical data only with external representatives, namely scores given to alternatives on each criterion, we argue that internal patterns of historical data are also essential to achieve a more comprehensive evaluation of alternatives. Nevertheless, a data-driven methodology aiming to capture internal patterns of historical data and consequently provide decision-makers with meaningful insights is still lacking. In this paper, we propose a framework that serves as integration within which both external representations and internal patterns are used to make a more comprehensive evaluation. Then we apply our framework to student evaluations and country-level risk assessments in the contexts of additive and fuzzy measures, respectively. Both cases show that either using external representatives or internal patterns alone leads to entirely different and even biased results. On the contrary, integrating external representations and internal patterns can offer more flexibility and higher interpretation power.