Ranking Multi Criteria Decision Making Methods for a Problem by Area Under Receiver Operating Characteristic

摘要

One of the major challenges in decision making is selection among MCDM (multi criteria decision making) methods. These methods do not provide same answer to decision maker. Therefore selecting the best answer is an important dilemma. To solve this problem, methods like Borda and Copeland compilation have been proposed. However, applying these methods leads to a hybrid solution which is not necessarily the best answer. In this paper a new approach is proposed to rank different MCDM methods. This approach is AUROC (area under receiver operating characteristic) which is a data mining tool for ranking classification models. The results would show great potential of applying AUROC for ranking MCDM methods in an immense selection problem with historical outcome.