A modified CRITIC with a reference point based on fuzzy logic and hamming distance

摘要

Multi-Criteria Decision Making (MCDM) aims to support decision-makers more effectively. This paper presents modifications for the criteria ranking methods, CRITIC and Distance-correlation CRITIC (D-CRITIC). These modifications involve the normalization procedure to silence criteria attribution by utilizing fuzzy logic and the hamming distance. Also, some adjustment to the weighting technique of the CRITIC method is done to capture linear and nonlinear correlations more precisely. These modifications that are applied to the CRITIC method by considering some reference point (RP), produced our proposed (M-CRITIC-RP) method. An illustrative example to rank performance criteria of display advertisement for a digital marketing campaign is used to demonstrate the method’s workability. Comparative analysis with appropriate MCDM methods reveals that the distance correlation values for the weights of the M-CRITIC-RP and comparative methods exhibit higher consistency than those between other criteria ranking methods. Moreover, M-CRITIC-RP has a high average value of Spearman rank statistic, indicating high rank consistency. Regarding accuracy, the Symmetric Mean Absolute Percentage Error (sMAPE) and the Log Accuracy Ratio (LAR) error metrics show lower values for M-CRITIC-RP in comparison to the corresponding values of other criteria ranking methods. M-CRITIC-RP demonstrates sensitivity with changes in the number of alternatives and also to variations in the RPs. The results of the sensitivity analysis guide the analyst to pay ample attention when selecting the RP. Findings also suggest considering a sufficient sample size for better accuracy and consistency of weights and ranks. Several implications were discussed along with advantages and limitations to be addressed for future work.