Correlation based sensitivity analysis and suitability evaluation of multicriteria decision-making methods according to different normalization techniques

Using multi-criteria decision-making (MCDM) methods and the most appropriate normalization techniques significantly affects the accuracy of the ranking results obtained.The study's primary purpose is to present new robust and practical evaluation strategies for both the suitability of normalization techniques and the sensitivity of MCDM methods.In this study, new strategies created with metrics different from those used in previous studies (Spearman correlation, mean absolute deviation and variation coefficient) are proposed to evaluate the suitability and sensitivity of nine different MCDM methods with seven different normalization techniques.Strategy 1 is presented among the proposed strategies to evaluate the suitability of normalization techniques, and Strategy 2 assesses MCDM methods' sensitivity.The most important advantage of the proposed strategies compared to other studies is they provide a more reliable and practical experience by testing the consistency of the results.The compatibility of the results obtained by applying the proposed strategies shows that they are dependable, practical, and robust.According to the effects of Strategy 1, the most suitable normalization technique for each examined MCDM method is the Linear normalization technique, whereas the most unsuitable technique is the Logarithmic normalization technique.According to the results of Strategy 2, the most sensitive methods affected by the change of normalization techniques are TOPSIS (The Order Preference by Similarity Ideal Solution) and CODAS (COmbinative Distance-based Assessment), and the least sensitive methods are COCOSO (Combined Compromise Solution) and VIKOR (VIseKriterijumska Optimizacija I Kompromisno Resenje).For the first time, more than one MCDM method was evaluated in terms of both the sensitivities of MCDM methods and the suitability of normalization techniques comparatively, and for this purpose, the new robust and practical strategies with reliable metrics (strategies 1 and 2) are presented.