#164
Dear Furkan Fahri ALTINTAŞ
I read your paper:
A Novel Method for Assessing the Weight Coefficients of Criteria within the Framework of Multi-Criteria Decision-Making: Measurement Relying on the Impacts of an Exponential Curve Function (MIEXCF)
My comments:
1- In the Introduccion you say “This method aims to assess and rank alternatives based on the preferences and priorities of decision-makers”
This is not strictly true because there are methods that are not based on preferences
2- Page 2 “The second characteristic relates to the distinctiveness or conflict among these criteria. By elucidating and leveraging these inherent characteristics embedded”
In reality this second characteristic is a consequence of the first, since criteria are ranked according to the information values obtained by the first characteristic.
3- Page 3 “Therefore, this study puts forth a novel method. The core aim of this research is to establish a framework that effectively captures the impact values among criteria through exponential curve functions, transforms these impact values into criteria weighting coefficients, and thereby produces dependable outcomes”
In my opinion you must first define the concept of impact. Normally, the impacting event alters in some way the imparted event, however, that impact determines criteria weighting coefficients? I doubt it, because I do not see what mechanism can produce that, and mainly, what for, if criteria weights do not participate in alternatives evaluation? However, it is different if in reality that process can register the interplay between criteria.
“Consequently, this study is perceived as a valuable addition to the existing body of literature concerning methods for computing criteria weighting coefficients, offering a fresh perspective to the field”
The interplay between criteria is old, the Simplex algorithm of Linear Programming, does it since 1948, and in fact, is a fundamental component of the Simplex, only that in here it is computed many times, according to the complexity of the problem and the number of iterations.
4- “Accordingly, the more disorder or distinctiveness a criterion exhibits compared to others, the greater its importance becomes”
I believe that disorder is one concept and distinctiveness is another, and opposite. If you have disorder, it is externally difficult to have a coherent answer, like in a telephone conversion where there is a lot of noise, or when one of the speakers has a poor domain of the language, the other speaker does not get a clear idea of what was said. This is the definition of entropy which maximum value is 1, and the information content 0, because the receiving speaker is left in the dark. He is confused because he could not understood what the other person said, there is a lot of possibilities, that is disorder
When the conversation is clear there si a very low entropy and a high information content in the conversation. This is what is significant, therefore, a criterion with high entropy is, from the point of view of communication, has a poor value
5- It is really interesting, and new for me, your explanation about the use of the exponential function in determining the impact of a criterion on another, as well working with a minimum and a maximum value. An impact happens when the variation of one variable causes a variation on another. However, which is the advantage of this procedure compared to lineal repression, where a variation of an independent variable produces changes in another dependent variable?
There is besides, another way to this by using the IF function, which is in Excel.
But what whatever the method, it is only a mathematical operation, and I wonder how do you know that a certain criterion affects another? For instance, you compute the influence that criterion environmental contamination has on criterion return on investment (IRR), but is there any relation between both criteria in real life? May or may be not.
Consequently, you can determine mathematically that influence, but in reality, it is inexistent; take as an example the construction of a fossil power plant; you can increase or decrease the CO2 contamination by the fumes spewed from the power house, but it does not affect the risk of the plant producing less energy, because they arer not related. In other words, you can compute an effect, but it may exist only in your mind.
If my assumption is correct, your method could be useful if you from the beginning know that dependencies exist. In my opinion, processing a MCDM matrix involves considering all criteria and alternatives simultaneously, something that is not done by any MCDM in existence, except by Linear Programming (LP).
Using SIMUS, a method based on LP you can learn about the changes produced by all criteria interrelationships.In this case, if you change a performance value (not weight) of a criterion, you can see quantitatively how close or far is the result for this criterion compared with its original target value. The target value may be anything, like money, manpower, percent of IRR, or number of jobs generation. For this, you need to work with resources, something that only PROMETHEE and LP allow
For instance, assume that the amount of CO2 allowed contamination is 25 cubic meters/ hour (a fictitious value), and that the IRR is 6 %. Suppose that even when the plant is spewing less that such contamination, thanks to costly equipment to treat exhaust, you want to reduce that cost by installing less equipment. You run your model and find for instance, that the target IRR has not changed, meaning that the reduction had no influence on it. However, it could have affected another criterion which is directly or indirectly related to environment, for instance, human health.
In addition, you are finally computing criteria weights that do not have any influence in a alternative’s evaluation.
These are my comments, I hope they can help you
Nolberto Munier