Appendix A. TOPSIS
Hwang and Yoon initially proposed TOPSIS (Technique for Order Preference by Similarity to Ideal Solution) in 1981. The goal of this strategy is to rank the alternatives by calculating the distance between each alternative and the positive- and negative-ideal solutions for decision-making issues and then determining the best option. The following steps describe the approach provided by [
66]:
Step 1: Decision matrix R = { }s, where (i = 1, 2, …, m; j = 1, 2, …, n) is the value of the th attribute in the th alternative, is identified in this step.
Step 2: The difference in attributes and the order of magnitude should be considered; then, decision matrix R is normalized, and the normalized matrix is transformed to = { .
Step 3: The weighted normalized decision matrices are found: = .
Step 4: and
are identified by the following equations:
Step 5: The relative closeness of each alternative is calculated in this step by the following equation:
The value of relative closeness reflects the relative superiority of the alternatives. larger indicates that alternative is better, whereas a smaller indicates that this alternative is poorer.
Appendix B. ELECTRE III
Let us imagine that we want to compare options and and determine which is the better alternative. ELECTRE is a self-contained procedure that does not require compensation. Procedures that are not compromise methods are frequently split into two halves. The first stage is outranking, and the second is exploitation. In the first step, we want to see if option is better than option. We attempt to figure out how they compare in the second step.
To accomplish this, we must first examine the ELECTRE family’s preference structure. The ELECTRE’s preference structure is denoted by the letters P, I, and R. The problem is that we aim to keep superiority apart from superiority. During the technique under consideration, we assert that choice
is superior to option
. When we make a decision, however, we use the word preference, which implies that option
is preferable to option
. This preference structure is depicted in Equations (A3)–(A6), which we explore. As stated in Equation (1), option
is superior to option
; however, option
is not superior to option
. As a result, option
is chosen over option
.
Equation (A4) shows that option
is superior to option
, and option
is not superior to option
. As a result, option
is strictly preferred to
.
Equation (A5) shows that both option
is superior to option
and option
is superior to option
, so option
and option
are equivalent or indifferent to each other.
Equation (A6) shows that option
is not superior to option
, nor is option
superior to option
. This relationship does not mean that the two options are indifferent to each other, but it does mean that the two options are not comparable.
When there is not enough information regarding an issue’s alternatives, we can add one or more criteria to the options. Adding one or more criteria to the mix may be enough to solve the problem.
ELECTRE should be used to solve problems that have at least 3 and no more than 13 criteria. It is also better if the criteria are varied or have specific numerical values. In ELECTRE, each criterion is compared to itself.
The ELECTRE approach is based on the calculation of two criteria. The declaration of these two conditions gives rise to the difference between the ELECTRE types. Let us assume that we want to evaluate options and and determine if option is better than option . We must consider the following two conditions:
Concordance. This condition tells us that to show the superiority of option over option, we must show the concordance between this pair. Let us show the evidence that supports the superiority of option over.
Non-discordance. In this condition, we gather evidence that indicates the superiority of option over.
These two conditions suggest that option must first meet the concordance criterion before being regarded as superior to option. In that instance, we have evidence that option is better than option . Second, there is no evidence that option is preferable to option . As a result, option is superior to option .
The aim is to have a total weight of one for all of the criteria.
Another factor to consider is the lack of veto power. In the non-discordance situation, we determine the disadvantages. When criteria are vetoed, if an option exceeds most of the requirements without a criterion being rejected, that criterion may veto the supremacy of all criteria. This is seen in Equation (A7).
As long as the difference in veto criteria between the two choices does not exceed, this relation shows that option A is preferable to option . is referred to as a veto limit in this context. The difference between options and should be less than the veto limit under this criterion. We use ELECTRE III to solve this problem since it considers uncertainty.
Let us say that we think option
is better than option
. We also suppose that
is variable and
is constant for criteria
, which is a profit measure. As indicated in
Figure A1,
represents our variable and the vertical axis of our discordance. Non-superiority is represented by a rejected alternative in this diagram. From
onwards, we declare complete non-superiority. That is, we combine the no-veto condition with the non-discordance condition. Equation (A8) represents the relationships in
Figure A1.
Figure A1.
Preference function in ELECTRE III.
Figure A1.
Preference function in ELECTRE III.
The bigger the
, as stated in Equation (A8), the larger the proportion. That is, our evidence of inferiority grows. The closer
is to
, the less we lack. Finally, we construct a parameter or variable called
, which is termed validity, to analyze the superiority of
over
. That is, Equation (A9) equals the validity of our assertion that option A is preferable to option
:
If the difference between options and in criterion exceeds a veto limit, it becomes and 1, which means that our claim that option is superior to has no validity.
If none of
is in the form of Equation (A10), this claim is eliminated, and we measure the value of this claim only by the value of concordance.
In fact, is the degree to which option claims superiority over option in the fuzzy set of preferences. The higher this value of , the higher the degree of belonging. This ELECTRE goes out of the range between zero and one because it takes into account uncertainty. When we say accept , we are defining a safe margin for .
Appendix C. VIKOR
The VIKOR suggested by Opricovic [
56] is explained here. As briefly mentioned previously, it focuses on ranking alternatives and determines compromise solutions for a problem with conflicting criteria.
Step 1. Construct the performance matrix and weight vector:
Additionally, and , where denotes the alternative represents the criterion (or attribute) ; indicates the fuzzy performance rating of alternative (district in this study) with respect to criterion (indicator in this study); and indicates the weight for each criterion.
Step 2. Determine the ideal
and the nadir
values of all criteria functions according to benefit or cost functions. The set of criteria representing benefits (good effects) is denoted by
, and the set
represents costs. (Equations (A11) and (A12)).
Step 3. Compute the normalized fuzzy difference
: (Equations (A13) and (A14)).
Step 4. Compute the values of
and
by the relations in Equations (A15) and (A16).
Step 5. Compute the values of
by the relation in Equation (A17).
where
and
. Additionally,
is introduced as a weight for the strategy of “the majority of criteria”
, whereas
is the weight of the individual regret
.
The weighting parameter is the maximum utility of a group; its value can be between 0 and 1 and is considered 0.5 in this research.
Step 6. Rank the alternatives, sorting in decreasing order. The results are three ranking lists , , and according to crisp(), crisp(), and crisp(), respectively.
Step 7. Propose a compromise solution the alternative , which is the best-ranked solution by the measure , if the following two conditions are satisfied:
In this step, we decide according to the , , and values of the options that are sorted in descending order. In order to decide, the following two conditions are considered:
C1. “Acceptable advantage”: .
Where is the advantage rate of alternative ranked first compared with the alternative with the second position in and the threshold
C2. “Acceptable stability in decision making”:
Alternative must also be the best ranked by or .
If one of the conditions is not satisfied, then a set of compromise solutions is proposed, which consists of:
CS1. Alternatives and if only condition is not satisfied, or
CS2. Alternatives ,, … , if condition is not satisfied; is determined by the relation for maximum . The positions of these alternatives are close.
Appendix D. PROMETHEE
PROMETHEE is a compromise strategy that uses the preference function to select the best option. It is used in a wide range of fields. This approach, in general, gives options and criteria, turns qualitative indicators into quantitative indicators, and weights the indicators so that the total weight equals one. The PROMETHEE II method is employed. Each option is given a numerical value in this method.
The preference function P is used to compare the two options
and
in the k index due to the distance between them. That is, it is dependent on the distance between the two possibilities, as shown in Equations (A18) and (A19).
Equation (A19) shows that the preference function
, which is for comparing the two options
and
in terms of the index
, is due to the distance between the two options. That is, it depends on the distance between options
and
. In this respect, D represents the distance. The distance between the value of option
and
and the value of option
is stepwise. The general form of this preference function is shown in
Table A1.
Table A1.
Preference function.
Table A1.
Preference function.
| | | … | | … | |
| | | … | | … | |
| | | … | | … | |
… | … | … | … | … | … | … |
| | | … | | … | |
… | … | … | … | … | … | … |
| | | … | | … | |
Suppose that
and
are two hypothetical options, and we denote the performance of option
for criterion
by
. Our dominance relationship between the two available options can be shown by one of the defined in Equations (A20)–(A22):
means complete superiority or mastery. One option has complete precedence or dominance over the other when, for each criterion, option
is better than
and there is a criterion called
in which criterion option
is strictly superior to option
.
Equation (A21) states that two options are equal when they are the same for each number in different criteria.
denotes incomparability in relation A15. We cannot determine which choice is better when option has absolute supremacy over option in a set of criteria, while option has absolute superiority over option in a set of criteria.
When option
is superior to option
, the magnitude of its superiority is not discernible from the aforementioned relationships. Preference functions are used in the PROMETHEE technique to eliminate this flaw and influence the intensity of the superiority of the choices. The larger the gap between the two possibilities in one criterion, the higher the degree of preference, according to several preference functions. The difference between the two alternatives with
in the
criteria is positive if the criterion is positive.
Figure A2 shows our preference function in this problem. Equation (A24) shows the relations of the preference function in
Figure A2.
is the degree of preference of over based on criterion .
is the distance between and based on criterion .
, and represent the indifference threshold, the preference threshold, and the midpoint between and , respectively.
Figure A2.
Preference function in PROMETHEE method.
Figure A2.
Preference function in PROMETHEE method.
To solve this problem, we use the function with the indifference threshold. As shown in the figure, we have a threshold such as
and
, which is a linear function.
The weight of criteria , the performance of each option in , and the preference function for each criterion must all be defined by the DM before employing the PROMETHEE techniques. Then, the next step is to simply follow the instructions below:
Step 1: Calculate the overall preference index of each option to the other options.
The overall preference index should be calculated for each pair of options according to Equations (A25) and (A26). The closer this index is to one, the stronger the overall preference of option
over
. Equation (A25) indicates the superiority of option
over option
.
There are a few aspects worth mentioning. Option
normally has no benefit over option
since the distance is zero, as indicated in Equation (A27). Depending on the value of the distance stated in Equation (A28), the benefit of option
over option
is anywhere between zero and one. Option
’s superiority over option
is also demonstrated in this way. In addition, the total of option
’s superiority over option
and option
’s superiority over option
is between zero and one, as indicated in Equation (A29).
Step 2: Calculate implicit dominance flows.
In this section, we calculate the positive and negative implicit dominance flows for each option. The positive implicit dominance flow is shown in Equation (A30).
It should be noted that
represents the sum of all options.
also represents the average degree of mastery of
over other options.
The negative implicit dominance flow equation is shown in Equation (A31).
indicates the average degree of mastery of the other options over
.
The lower is, and the higher is, and the better the choice.
Step 3: Perform ranking with PROMETHEE II.
In the PROMETHEE II method, using the
and
interface and the results obtained from the first and second steps, a complete ranking is obtained in the form of Equations (A32) and (A33).
(
) or net flow is the implicit dominance of option
, which is represented by Equation (A34).
In general, PROMETHEE II requires calculating for each of the options and then subtracting the advantage from the failure to achieve a net dominance. The higher this, the better. In this method, we have a final and complete ranking, and we can determine the position of each option.