AbstractThis Dissertation concerns helping people to make complex decisions. A new approach to select a Multiple Criteria Decision Making (MCDM) method for a problem is created. The approach is based on an analysis of MCDM problems and methods and robustness of results. That revealed some factors to be addressed when selecting a method, including problem and method characteristics.
The new approach suggests a group of candidate methods suitable for a specific problem, then conducts sensitivity analysis on the outcomes of the candidate methods to recommend the method that delivered the most robust outcome. Sensitivity analysis is used for the first time to provide a scientific justification for selecting a MCDM method. Once performance measures and criteria weights are identified then sensitivity analysis and other mathematical approaches are applied to the output of the methods to define the most critical criteria and the most critical performance measures. A MCDM method is recommended that provides the most robust outcome with a compromise between the minimum percentage change required in the most critical criteria and the most critical performance measures to alter the outcome.
A new MCDM Methods Recommendation Model (MRM) was created to improve the decision making process by introducing a Reliability Threshold, Credibility Threshold and Credibility Value. Suitable methods are then recommended for a problem based on the robustness of their outcomes to risk and uncertainty anticipated in criteria weights.
A set of new Propositions, Rules and Hypotheses are presented based on potential generalized problems. These are tested on a number of numerical examples and revisions are made.
The new approach is successfully applied to real world problems, including human resource management, strategic marketing and corporate relocation. Finally the Preference Ranking Organization METHod for Enrichment of Evaluations II (PROMETHEE II) method is applied to suggest a steering direction for powered wheelchairs as a first step towards future work.
The work did not compare the outcome of MCDM methods but compared the stability of the outcome for a specific problem when uncertainty affected both criteria weights and performance measures to recommend a method that delivered the most robust outcome. The new approach, model and set of Propositions, Rules and Hypotheses aim to help decision makers in making complex decisions.
|Date of Award||May 2019|
|Supervisor||David Sanders (Supervisor), Giles Tewkesbury (Supervisor) & Nils Bausch (Supervisor)|