# 149
Dear Do Duc Trung, Tran Van Dua
I have read your article
Development of RAMDOE: a new method for rapidly ranking alternatives with supplementary options and considering changes in criteria values
My comments
1- The title really called my attention because this is a theme I have never seen addressed before, and I find it a very interesting subject, that appears often in DOE.
What you say is true, if you need to change alternatives and/or criteria it can be a very cumbersome and time-consuming task.
If you allow me, I would like to mention how I solved it in a different manner, using SIMUS, just by building an initial matrix with data and with surplus empty columns (alternatives) and surplus empty rows (criteria). I can run the matrix with a certain structure and data without any problem. The use of surplus columns and rows simplifies enormously the problem, without altering the initial data, and practically taking no extra time
a) If there is a need to eliminate an option just go to the original matrix and delete the column
b) If there is a need to add one or more alternatives either independently or at the same time, just use one or several of the empty columns, al put the new value for each criterion
c) If there is a need to correct only some performance values, delete it and add the new one
d) Exactly the same procedure for criteria
e) These operations can be done at the same time for alternatives and criteria, i.e., it is possible to work changing alternatives and criteria at the same time.
f) If a new criterion calls for cost while all theirs call for benefits. Express that with a new criterion
g) If you need to use simulation, just instruct Montecarlo to do that on the corrected matrix.
In this way, you can use as many modifications as you want in a short period and have each run saved.
Sorry for my lengthy precedent paragraphs, but I thought they may interest you and other colleagues with this problem. Of course, if you are interested it will be my pleasure to share it with you and with any other researcher or practitioner. ([email protected])
If by a chance you have the SIMUS software, I suggest to verify this, if not, let me know I and I will sent you a zip file with the software, from which you can make many copies and share it at will.
2- In page 1 you say” allexcelling in identifying the optimal choice among alternatives”
In MCDM there are not optimal choices. All methods look for a balance or a compromise solution
3- “However, using traditional MCDM methods alone requires recalculating the entire process when the number of alternatives changes (due to additions or removals)”
True, but it is the computer that does the calculations and it can take only no more that 2 or 3 minutes.
4- “However, this study has not considered the case where the values of the criteria in the additional option fall out[1]side the range of values of the criteria in the existing options”
This is difficult for me to understand. If you normalize all criteria, all of their values will be between 0 and 1
5- “This is perhaps something that the authors of this study have not fully anticipated, that there may be additional options whose criterion j does not fall within the range from x to y.”
Not, if you normalize. You mention this fact several times, as if in criteria there could be outliers. We are in MCDM, not in regression analysis. We are not looking for average values but , we work with linear equations that could or not be related
6- “To address this limitation, this study proposes a solution to rank alternatives considering the adjustment of criterion values in the additional alternatives”
In my opinion, when you add an alternative, you are adding its column of criteria values, that is, one value for criterion, and consequently, they forcefully will affect the attribute or content of existent attribute in each criterion, since now the number of performance values changed. This may or not change the existing dispersion values within each criterion, and thus, it has a direct relationship con its capacity to evaluate alternatives, according to Shannon’s entropy. Remember that attributes are the characteristics of the values within each criterion.
7- In page 3 where did you get formula (6) from? You do noy say how the weights are calculated
8- ” Step 3: construct the table of limit values of the criteria after adjusting their values as in Table 1”.
In general, in MCDM limits for criteria minimums and maximums are considered targets that need to be reached. For instance, for a criterion like ‘water supply’, it is normal to establish a minimum amount of litters/person-day and a maximum value to avoid waste
These are my comments, I hope they can help
Nolberto Munier