Dear Jakub Wieckowski , Zdzisław Szyjewskib
I have read your paper:
Practical Study of Selected Multi-Criteria Methods Comparison
Here are some of my comments
1 You say “It can be seen that the MULTIMOORA and MOOSRA methods provided equal rankings”
Fine. Does it mean that coincidence of rankings is a solid and sufficient reason to consider them as a reference or yardstick used for comparison? Is there any axiom or theorem that justifies this assumption? I am not judging any method, only asking a question, because I don’t understand how two similar or even different methods are taken as a yardstick.
2- In page 1 you say “However, the resulting rankings should be as consistent and reliable as possible”
True, if we start with the same data, aiming at the same result and using mathematical algorithms, the ranking should be, if not equal, at least similar, but unfortunately, it does not normally happen.
However, even if the results coincide, it does not mean that it is the real result, simply because we never have something to compare to.
3- In page 2 “For this purpose, they use a set of criteria and corresponding weights that determine their relevance to the problem”
I am afraid I disagree. Weights don’t determine the relevance of criteria to the problem, except when using objective weights. Subjective weights ‘determine’ or in reality, assign arbitrarily, a relative value of importance between criteria, no more than that.
As a proof of this, think that weights obtained from AHP for instance, are assigned WITHOUT considering alternatives. Therefore, apparently, as is done at present, it is the same to compare criteria Cost and Environment, and arbitrarily say for instance that cost is 3 times more important than environment, either for a school construction or for a road construction. It does not seem very rational.
“This approach allows for a comprehensive problem analysis to ensure reliable results”
Where are the analysis and the reliable results?
“and their effectiveness has been repeatedly verified “
How?
4- On Page 3 “To indicate which techniques have a high similarity of results, appropriate measurement measures must be used”
And where is the gain in comparing rankings?
As a real and common example. Assume you must travel between cities X and Y, 750 km apart, by road, which is covered by five bus companies. You have many criteria like cost, travel time, comfort, frequency, attention on board, etc. You use six different MCDM methods, and three of them, A, D and F, have very similar rankings and also high correlation, taking pairs.
Now, what did you learn from this computation? Nothing
Did you solve your problem of selecting the best equipment? Obviously not
Did the good consistence od rankings help you? No.
Then you are at square one, although with less companies to choose from.
Last question: Which was the preferred laptop which was the purpose of this study?
I hope that these few comments may help you.
Nolberto Munier
- Badges
- Science topic
- Similar topics
- Statistics
- Survey