Multi-dimensional stability analysis for Analytic Network Process models

The decision-making process involves, most of the time, working with limited and/or incomplete information to obtain a one-time synthesis of one's preferences. However, additional information might be acquired by the decision-maker at a later time. To identify the impact that the additional information could have on the initial result, we developed a new multi-dimensional stability analysis that provides a comprehensive image of how preferences change or evolve, as perturbations are applied to the criteria weights, without reiterating the preference elicitation process. The insights provided by the new method helped define a set of stability measures useful in a practical setting. To show the direct implementation of the findings of the multi-dimensional stability analysis, we applied the method developed to a randomly generated ANP model.

Automated summary

The decision-making process often involves limited and incomplete information. In order to understand the impact of additional information on the initial result, a new multi-dimensional stability analysis method was developed. This method allows for the observation of preference changes as criteria weights are perturbed, providing useful stability measures.