Selection of a suitable MCDA method based on robustness of results and sensitivity analysis

Making a decision is a process where alternatives are assessed to select a choice or a course of action to fulfill desired objectives and goals. A suitable decision making process is essential for success in an organization. Unsuitable and deficient decision-making might reduce competitiveness. Multiple-Criteria Decision-Making (MCDM) has often been considered as a reliable method for decision-making. MCDM is a set of methods and procedures by which multiple and conflicting criteria can be incorporated into the decision process.

MCDM aim is to enable decision makers to solve conflicting real world quantitative and or qualitative multi-criteria problems, and to find best-fit alternatives from a set of alternatives in certain, uncertain, or risky environments.

A new framework to select an appropriate group of candidate MCDM methods for a decisional problem is being created in this research. A new structured framework is proposed based on an analysis of MCDM problems and methods. That research is revealing factors to be addressed when selecting a MCDM method, including problem characteristics and MCDM method characteristics. Mathematical approaches are being applied to candidate methods to select the method that will provide the most robust output.

Problem characteristics considered in the framework will address the nature of alternative sets (continuous or discrete), type of input set (qualitative, quantitative, or mixed), the nature of the information being considered (deterministic, non-deterministic or mixed), the type of decision problem being addressed (choice, ranking, description or sorting) and the type of preference mode being considered (pairwise comparisons or performance measures). Moreover MCDM methods characteristics will address the type of ordering of alternatives (total order, partial order or interval), the measurement scale (nominal, ordinal, interval or absolute), and the type of preference structure (preference, indifference or incomparability), software availability and ease of use (ease of understanding, user friendliness, previous experience and knowledge, and time needed to apply a method). By addressing these factors, this framework can provide decision makers with a suggested group of candidate MCDM methods that are appropriate for their MCDM problem.

Finally sensitivity analysis and mathematical approaches will be applied to the suggested group of candidate methods to automatically select a MCDM method that delivers the most robust outcome. This will be achieved by calculating the minimum percentage change in criteria weights and performance measures required to reverse the ranking of any two alternatives.