|
Sources | Basic specifications | Positives (P)/negatives (N) |
|
[15] | Direct extension of Saaty’s AHP model with triangular fuzzy numbers | (P) The judgments of multiple experts may be modeled in the reciprocal matrix |
Lootsma’s logarithmic least square model is used to derive fuzzy weights and fuzzy performance scores | (N) Evermore there is no solution to the linear equations |
(N) The calculational demand is great, even for a low problem |
(N) It permits only triangular FN to be applied |
|
[16] | Direct extension of Saaty’s AHP model with trapezoidal fuzzy numbers | (P) It is simple to extend to the fuzzy term |
Uses the geometric mean model to derive fuzzy weights and performance scores | (P) It guarantees an alone solution to the reciprocal comparison matrix |
(N) The calculational demand is great |
|
[33] | Modifies van Laarhoven and Pedrycz’s model | (P) The judgments of multiple experts may be modeled |
Shows a more robust method to the normalization of the local priorities | (N) The calculational demand is great |
|
[32] | Synthetical degree values | (P) The calculational demand is partly low |
Layer simple sequencing | (P) It follows the steps of crisp AHP; it does not involve additional process |
Composite total sequencing | (N) It permits only triangular FN to be applied |
|
[32] | Creates fuzzy standards | (P) The calculational demand is not great |
Illustrates performance numerals by membership functions |
(N) Entropy is applied when probability distribution is known. The model is based on both probability and possibility measures |
Uses entropy concepts to calculate aggregate weights | |
|
Proposed method | Uses interval approximation to defuzzification | (P) It permits all FN to be applied |
(P) It retains uncertainty of fuzzy number in itself |
Uses goal programming method to obtain weights | (P) It is done, mainly, with a software such as Lingo |
|