What is the formal rationale for the specific ratios 0.55:0.33:0.11? Are they derived from a normalization of a fixed scale (e.g., 5:3:1)? If so, what justifies this scale?
If these weights are meant to be customizable by the DM, shouldn’t their elicitation be a fundamental step in the SCR protocol? Their current treatment as a fixed input, separate from the DM’s membership assessments (μxj), creates a disjointed process where the final weights are partly subjective (via μxj) and partly based on an unexplained, fixed constant.
Have you tested the sensitivity of the final criteria weights (wj) and the resulting alternative rankings to variations in these cluster weights (wα,wθ,wβ)? A small change in these base weights could propagate and alter the final rankings, which would question the robustness of the method.
1. Figure 18 claims to show a positive correlation between SCR and Entropy weights, but visual inspection suggests the correlation is weak at best (points appear widely scattered). The paper doesn’t provide the actual correlation coefficient value. What is the Pearson correlation coefficient, and is it statistically significant? Additionally, the Manhattan Distance results in Table 11 show Entropy weights differ substantially from SCR weights (0.092 vs. 0.027 for CRITIC), contradicting the claim of strong similarity.
2. In the paper, Figure 15 and Figure 16 both have captions stating Three abstract rectangles architected on a creation membership values – this appears to be a typo where “criterion” was intended. More importantly, the figures referenced as Figures 11-16 show abstract rectangles, but the paper never clearly explains which specific membership values (l, e, or u) serve as the width and height for each rectangle in the actual weight calculation algorithm.