# 131

I read your paper

Data Normalization for Root Assessment Methodology

My comments:

1- In page 1 you say “According to the proponent of the RAM method, this method has the advantage of overcoming the shortcomings of existing MCDM”

This is a too general assertion, because you did not say which are the shortcomings of the existing MCDM methods, that are several, and depending of the method

2- “The advantage is the ability to compensate between beneficial criteria and non-beneficial criteria”

And why should there be need for compensation? Benefits criteria as well as non-beneficial criteria are part of a problem.

3- Limiting rank reversal?

How? If nobody knows what RR is, how come hat RAM can put a limit to it?

4- Ina page 2 “, but adding a solution that is worse than the previous solutions cause the ranks of the previous solutions to change”

And how do you know that the added alternative – not a solution, which are the results of the analysis – is worse than other alternatives? Just by observing the values of the corresponding new vector? Hardly, because the new vector will modify the existing initial matrix and you cannot foresee, let alone compute, how this vector will interact with the others. In addition, common sense says that if the alternative is worse, why is the purpose in adding it?

5- Consider another situation, “if a certain criterion is of type C (the smaller the better) and its value in a certain solution is zero, then methods N1, N3, N5, N7, N10 also cannot be used”

There is misconception regarding the maximum the better and the smaller the better.

In the case of smaller the better consider for instance a criterion like ‘Environmental contamination’; you must consider that zero pollution normally does not exist, because everything produces contamination, even your breath. If you fabricate a car, of course that you must aim at the minimum cost, but it can never be zero, it would be irreal, because you cannot fabricate something at zero cost of any nature, like effort, space, money, tie, etc.

Similar happens in maximizing, yes, the maximum the better, but it has a limit, since you are able to produce or build something subject to the number of resources you have, therefore, the maximum will be limited by them. For instance, in the marine shipping business, of course, the more the load the greater the benefit, but the load is limited by the ship’s capacity and design.

In other words, if you do not consider the available resources, your calculation using MCDM is faulty.

6- You refer at mixing normalization methods something that I disagree. I cannot understand why it is different ranking telephones than robots. Of course, both require different set of criteria. And I do not understand either what you say about customers’ feedback. If you have a criterion named ‘Maximize positive feedbacks” and another named ‘Minimize negative feed backs”, and if you use for both the same scale, say 1 to10, the maximum the better, you can use the same scale for negative feedbacks. Since in the first case you maximize, the maximum will be the better. In the second case the maximum will be the worse

I hope these comments may help you

Nolberto Munier