Abstract
We define \(2n+1\) and 2n fuzzy numbers, which generalize triangular and trapezoidal fuzzy numbers, respectively. Then, we extend the fuzzy preference relation and relative preference relation to rank \(2n+1\) and 2n fuzzy numbers. When the data is representable in terms of \(2n+1\) fuzzy number, we generalize the FMCDM (fuzzy multi-criteria decision making) model constructed with TOPSIS and relative preference relation. Lastly, we give an example from telecommunications to present the proposed FMCDM model and validate the results obtained.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
Zadeh (1965) introduced the concept of fuzzy set, and it is widely used to characterize vague or imprecise settings (conditions). Fuzzy sets have applications in automata theory, systems theory, decision theory, switching theory, pattern recognition, image thresholding, etc. [Lalotra and Singh (2020); Singh et al. (2019); Singh and Sharma (2019); Singh et al. (2020)]. Fuzzy numbers generalize real numbers and are very useful to represent data corresponding to uncertain situations. There are several methods to rank or order fuzzy numbers. Lee and Li (1988) utilized the concept of probability measure to determine the order of fuzzy numbers by considering the mean and dispersion of alternatives. Choobineh and Li (1993) proposed an indexing method to order or rank the fuzzy numbers. Dias (1993) proposed a computational approach to rank the alternatives using fuzzy numbers. Fortemps and Roubens (1996) presented a method to compare fuzzy numbers using the area compensation procedure. Cheng (1998) proposed the distance method and coefficient of variation (CV) index method to rank the fuzzy numbers. Chu and Tsao (2002) proposed a method using the area between centroid point and original point of the fuzzy numbers to facilitate ranking. Wang and Lee (2008) later revised this method. Lee (2005b) introduced the ’comparable’ property for fuzzy preference relation and showed that only O(n) comparisons of fuzzy numbers are sufficient if a fuzzy preference relation satisfies certain conditions. Asady and Zendehnam (2007) proposed a ranking method for the fuzzy numbers by obtaining the nearest point of support function with respect to fuzzy quantity. Wang (2015b) proposed a fuzzy relation with membership function representing preference degree to compare two fuzzy numbers. A relative preference relation was defined using fuzzy preference relation to compare a set of fuzzy numbers. The relative preference relation expresses preference degrees of several fuzzy numbers over average and facilitates easy and quick ranking of fuzzy numbers.
Decision-making methods often apply fuzzy sets in their computations. Jain (1976) presented a decision method that represented uncertain quantities as fuzzy sets and subsequently obtained an optimal alternative. Jain (1977) also developed a procedure for decision making using fuzzy sets by assigning quantitative numbers to qualitative terms. Wang (2014, 2015a, 2020a, 2020b) proposed various methods using relative preference relation to solve FMCDM problems. In the multi-granulation decision-theoretic rough set, Mandal and Ranadive (2019) introduced the optimistic and pessimistic fuzzy preference relation models.
In a multi-criteria decision-making problem with multiple data points, few data points, referred to as fuzzy numbers, are utilized to arrive at a decision. In this regard, different fuzzy numbers, such as triangular, trapezoidal, pentagonal and hexagonal fuzzy numbers, have been reported. These fuzzy numbers consider only a few data points to arrive at a decision. For example, the number of data points in triangular fuzzy numbers is 3, in trapezoidal fuzzy numbers it is 4, and in hexagonal fuzzy numbers it is 6. However, using a few data points to represent data leads to the loss of information. To address this situation, we generalize fuzzy numbers that encompass more data points to represent the data, thus minimizing the loss of information. Practically, when the decision problem is highly sensitive to the number of data points, it is reasonable to choose a larger value of n. The flexibility in implementing this idea is apparent in the case of data representation by \(2n+1\) (or 2n) fuzzy numbers due to the choice for n. Thus, we get a natural advancement to the existing FMCDM methods.
In this paper, we define 2n and \(2n+1\) fuzzy numbers. Clearly, \(2n+1\) fuzzy numbers yield triangular and pentagonal cases when \(n=1,2,\) and 2n fuzzy numbers coincide with trapezoidal and hexagonal fuzzy numbers when \(n=2,3\), respectively. We extend the fuzzy preference relation and relative preference relation given by Wang (2015b) to rank 2n and \(2n+1\) fuzzy numbers. Then, we compare the results obtained by fuzzy preference relation and relative preference relation with Wang and Lee (2008) method. Wang (2014) developed the FMCDM model with TOPSIS under fuzzy environment and relative preference relation on fuzzy numbers. We present an extension to the FMCDM model when the given data is representable in terms of \(2n+1\) fuzzy numbers. We illustrate the suitability of the proposed method in solving FMCDM problems using an example. Subsequently, the proposed method results are validated and compared with VIKOR, MOORA and ELECTRE methods.
The rest of the paper is organized as follows. In Sect. 2, we present the basic definitions and related primary results. In Sect. 3, we give the definition of \(2n+1\) fuzzy number and the extension of fuzzy preference and relative preference relations on \(2n+1\) fuzzy numbers. In Sect. 4, we define 2n fuzzy number and the extension of fuzzy preference and relative preference relations on 2n fuzzy numbers. In Sect. 5, we present the proposed FMCDM model along with a telecommunication example. In Sect. 6, we validate the proposed method with popularly used multi-criteria decision-making methods.
2 Definitions and preliminaries
For the following definitions, we refer (Zadeh 1965; Zimmermann 1987, 1991).
Definition 2.1
A fuzzy subset A on the universe U is a set defined by a membership function \(\mu _{A}\) representing a mapping \(\mu _{A}:U\longrightarrow [0,1].\)
Definition 2.2
\(A_{\alpha }=\{x\mid \mu _{A}(x)\ge \alpha \}\) is called an \(\alpha \)-cut of the fuzzy set A.
Definition 2.3
Let X be a fuzzy number. Then, \(X_{\alpha }^{L}\) and \(X_{\alpha }^{U}\) are, respectively, defined as
Definition 2.4
(Lee 2005a, b; Epp 1990) A fuzzy preference relation R is a fuzzy subset of \({\mathbb {R}} \times {\mathbb {R}}\) with membership function \(\mu _{R}(A, B)\) representing preference degree of fuzzy numbers A over B.
-
1.
R is reciprocal if and only if \(\mu _{R}(A, B)=1-\mu _{R}(B, A)\) for all fuzzy numbers A and B.
-
2.
R is transitive if and only if \(\mu _{R}(A, B)\ge \frac{1}{2}\) and \(\mu _{R}(B, C)\ge \frac{1}{2}\Rightarrow \mu _{R}(A, C)\ge \frac{1}{2}\) for all fuzzy numbers A, B and C.
-
3.
R is a fuzzy total ordering if and only if R is both reciprocal and transitive.
A is preferred to B if and only if \(\mu _{R}(A, B)>\frac{1}{2}\) and A is equal to B if and only if \(\mu _{R}(A, B)=\frac{1}{2}.\)
Definition 2.5
(Wang 2015b) Let \(\succ \) be a binary relation on fuzzy numbers defined by \(A \succ B\) if and only if A is preferred to B (That is, \(\mu _{R}(A, B)>\frac{1}{2}\)).
Wang revised the extended fuzzy preference relation defined by Lee (2005b) as follows.
Definition 2.6
(Wang 2015b) Let A and B be two fuzzy numbers, where A is an interval \([a_{l}, a_{r}]\) and B is an interval \([b_{l}, b_{r}].\) A fuzzy preference relation P is a subset of \({\mathbb {R}}\times {\mathbb {R}}\) with membership function \(\mu _{P}(A, B)\) representing preference degree of A over B.
Define
where
\(T^{+}\) is an interval \([t_{l}^{+},t_{r}^{+}],T^{-}\) is an interval \([t_{l}^{-},t_{r}^{-}],t^{+}_{l}=max\{a_{l},b_{l}\},\) \(t^{+}_{r}=max\{a_{r},b_{r}\},t^{-}_{l}=min\{a_{l},b_{l}\}\) and \(t^{-}_{r}=min\{a_{r},b_{r}\}.\)
Similarly, fuzzy preference relation on triangular and trapezoidal fuzzy numbers have also been defined.
For the examples on decision-making problems, we refer (Koppula et al. 2019, 2020; Riaz et al. 2020; Chen and Huang 2021).
Definition 2.7
(Wang 2015b) Let \(S=\{X_{1}, X_{2},\ldots ,X_{n}\}\) denote a set composed of n fuzzy numbers. A fuzzy number \(X_{i}=[x_{il}, x_{ir}]\) belongs to the set S, where \(i=1,2,\ldots ,n.\) Assume \({\overline{X}}=\frac{\sum _{i}X_{i}}{n}\) derived by extension principle is average of the n fuzzy numbers in S. A relative preference relation \(P^{*}\) with membership function \(\mu _{P^{*}}(X_{i},{\overline{X}})\) represents preference degree of \(X_{i}\) over \({\overline{X}}\) in S.
We define
where
\(T^{+}_{s}\) is an interval \([t_{sl}^{+},t_{sr}^{+}],T^{-}_{s}\) is an interval \([t_{sl}^{-},t_{sr}^{-}],\) \(t^{+}_{sl}=\max \nolimits _{i}\{X_{il}\},\) \(t^{+}_{sr}=\max \nolimits _{i}\{X_{ir}\},\) \(t^{-}_{sl}=\min \nolimits _{i}\{X_{il}\}\) and \(t^{-}_{sr}=\min \nolimits _{i}\{X_{ir}\}.\)
Clearly, \(0< \mu _{P_{*}}(X_{i}, {\overline{X}})<1,\) where \(i=1,2,\ldots ,n\). \(\mu _{P_{*}}(X_{i}, {\overline{X}})<\frac{1}{2}\) expresses that \({\overline{X}}\) is preferred to \(X_{i}.\) On the other hand, \(\mu _{P_{*}}(X_{i}, {\overline{X}})>\frac{1}{2}\) expresses that \(X_{i}\) is preferred to \({\overline{X}}.\)
Similarly, relative preference relation is defined on triangular and trapezoidal fuzzy numbers.
3 Generalized \(2n+1\) fuzzy number
3.1 Generalized linear \(2n+1\) fuzzy number
Let \(\{a_{1}, a_{2},a_{3},\ldots ,a_{2n+1}\}\) be real numbers such that \(a_{1}<a_{2}<a_{3}<\cdots <a_{2n+1},~n=1,2,3,...\) and n is finite, \(k\ge 2^{n-1}.\) Then, we denote
Now, \([P(x), Q_{1}(x)]\) gives fuzzy membership function of the generalized triangular fuzzy number \((a_{1}, a_{2}, a_{3};\dfrac{1}{k}).\)
That is,
If \(k=1\) in the above, we get a triangular fuzzy number.
Now,
gives fuzzy membership function of the generalized fuzzy number \((a_{1}, a_{2}, a_{3},\ldots ,a_{2n+1};\dfrac{2^{n-1}}{k}),\) where \(n\ge 2\) and \(k\ge 2^{n-1}.\)
In particular, if \(k=2^{n-1}\) then (1) gives fuzzy membership function of the fuzzy number \((a_{1}, a_{2}, a_{3},\ldots ,a_{2n+1};1).\)
For example, substitute \(n=2\) in (1), then \([P(x), S_{2}(x), T_{(2,1)}(x), Q_{2}(x)]\) gives fuzzy membership function of the generalized pentagonal fuzzy number \((a_{1}, a_{2}, a_{3}, a_{4}, a_{5};\dfrac{2}{k}).\) That is,
If \(k=2\) in the above, we get a pentagonal fuzzy number (Fig. 1).
Similarly, substitute \(n=3\) in (1) then \([P(x), S_{2}(x), S_{3}(x), T_{(3,1)}(x), T_{(4,2)}(x), Q_{3}(x)]\) gives fuzzy membership function of the generalized heptagonal fuzzy number \((a_{1}, a_{2}, a_{3}, a_{4}, a_{5}, a_{6}, a_{7};\dfrac{4}{k}).\)
That is,
If \(k=4\) in the above, we get a heptagonal fuzzy number.
3.2 Generalized nonlinear \(2n+1\) fuzzy number
Let \(\{a_{1}, a_{2},a_{3},\ldots ,a_{2n+1}\}\) be real numbers such that \(a_{1}<a_{2}<a_{3}<\cdots <a_{2n+1},~m,n=1,2,3,...\) and m, n are finite. Now, take \(k\ge 2^{n-1}.\) Then, we denote
Then, \([P^{m}(x), Q_{1}^{m}(x)]\) gives fuzzy membership function of the generalized nonlinear triangular fuzzy number \((a_{1}, a_{2}, a_{3};\dfrac{1}{k})\). Now,
gives fuzzy membership function of the generalized nonlinear fuzzy number \((a_{1}, a_{2}, a_{3},\ldots ,a_{2n+1};\dfrac{2^{n-1}}{k}),\) where \(n\ge 2\) and \(k\ge 2^{n-1}.\)
In particular, if \(k=2^{n-1}\) then (3) gives fuzzy membership function of the nonlinear fuzzy number \((a_{1}, a_{2}, a_{3},\ldots ,a_{2n+1};1).\)
Note 3.3
If \(m=1\) in the fuzzy membership function of the generalized nonlinear \(2n+1\) fuzzy number, then we get a fuzzy membership function of the generalized linear \(2n+1\) fuzzy number.
3.3 \(\alpha \)-cut of a Generalized linear \(2n+1\) fuzzy number
Let \(\{a_{1}, a_{2},a_{3},\ldots ,a_{2n+1}\}\) be real numbers such that \(a_{1}<a_{2}<a_{3}<\cdots <a_{2n+1},~n=1,2,3,...\) and n is finite. Now, take \(k\ge 2^{n-1}.\) Then, we denote
Then, \([E(\alpha ), H_{1}(\alpha )]\) gives \(\alpha \)-cut of the generalized triangular fuzzy number \((a_{1}, a_{2}, a_{3};\dfrac{1}{k}).\) That is,
If \(k=1\) in the above, we get an \(\alpha \)-cut of a triangular fuzzy number.
Now,
gives \(\alpha \)-cut of the generalized fuzzy number \((a_{1}, a_{2}, a_{3},\ldots ,a_{2n+1};\dfrac{2^{n-1}}{k}),\) where \(n\ge 2\) and \(k\ge 2^{n-1}.\)
For example, if \(n=3\) in (5) then \([E(\alpha ), F_{2}(\alpha ), F_{3}(\alpha ), G_{(3,1)}(\alpha ), G_{(4,2)}(\alpha ), H_{3}(\alpha )]\) gives \(\alpha \)-cut of the generalized heptagonal fuzzy number \((a_{1}, a_{2}, a_{3}, a_{4}, a_{5},a_{6},a_{7};\dfrac{4}{k}).\) That is,
If \(k=4\) (That is \(k=2^{n-1}=2^{3-1}\)) in the above, we get an \(\alpha \)-cut of a heptagonal fuzzy number \((a_{1}, a_{2}, a_{3}, a_{4}, a_{5},a_{6},a_{7}).\)
Infimum and Supremum of \(\alpha \) -cut of a \(2n+1\) fuzzy number:
Let B be any fuzzy number and \(\mu _{B}(x)\) is the fuzzy membership function of B. Then \(B_{\alpha }^{L}=\inf \nolimits _{\mu _{B}(x)\ge \alpha }(x)\) and \(B_{\alpha }^{U}=\sup \nolimits _{\mu _{B}(x)\ge \alpha }(x)\).
For example, if \(B=(a_{1}, a_{2}, a_{3};\dfrac{1}{k})\) is a triangular fuzzy number. Then, \(B_{\alpha }^{L}=E(\alpha )\) and \(B_{\alpha }^{U}=H_{1}(\alpha ).\)
For a fuzzy number \(C=(a_{1}, a_{2}, a_{3},\ldots ,a_{2n+1};\dfrac{2^{n-1}}{k}),\) \(C_{\alpha }^{L}=(E(\alpha ),F_{2}(\alpha ),F_{3}(\alpha ),\ldots , F_{n-1}(\alpha ),F_{n}(\alpha ))\) and
3.4 Extension of fuzzy preference relation on \(2n+1\) fuzzy numbers
We extend the fuzzy preference relation given by Wang (2015b) to rank 2n+1 fuzzy numbers as follows.
Definition 3.6
Let \(X_{1}\) and \(X_{2}\) be two fuzzy numbers, where \(X_{1}=[a_{1},a_{2},\ldots ,a_{2n+1}]\) and \(X_{2}=[b_{1},b_{2},\ldots ,b_{2n+1}]\). An extended fuzzy preference relation R is a subset of \({\mathbb {R}}\times {\mathbb {R}}\) with membership function \(\mu _{R}(X_{1}, X_{2})\) representing preference degree of \(X_{1}\) over \(X_{2}\). Then,
where
\(T^{+}\) is an interval \([t_{1}^{+}, t_{2}^{+},\ldots ,t_{2n+1}^{+}],\) \(T^{-}\) is an interval \([t_{1}^{-}, t_{2}^{-},\ldots ,t_{2n+1}^{-}]\) and \(t_{1}^{+}\)=max \(\{a_{1}, b_{1}\}\), \(t_{2}^{+}\)=max \(\{a_{2}, b_{2}\}, \ldots ,t_{2n+1}^{+}\)=max\(\{a_{2n+1}, b_{2n+1}\},\) \(t_{1}^{-}\)=min \(\{a_{1}, b_{1}\}\), \(t_{2}^{-}\)=min \(\{a_{2},b_{2}\},\)...,\(t_{2n+1}^{-}\)=min \(\{a_{2n+1}, b_{2n+1}\}.\)
In (I), if \(\int _{0}^{\frac{2^{n-1}}{k}}({(T^{+}-T^{-})^{\alpha }_{L}+(T^{+}-T^{-}) ^{\alpha }_{U}})d\alpha \ge 0,\) then \(\mu _{R}(X_{1}, X_{2})\ge \frac{1}{2}\) and if \(X_{1}=X_{2},\) then \(\mu _{R}(X_{1}, X_{2})=\frac{1}{2}.\)
Lemma 3.7
The extended fuzzy preference relation R is reciprocal. That is, \(\mu _{R}(X_{1}, X_{2})=1-\mu _{R}(X_{2}, X_{1})\) for all \(2n+1\) fuzzy numbers \(X_{1}\) and \(X_{2}.\)
Lemma 3.8
The extended fuzzy preference relation R is transitive. That is, if \(\mu _{R}(X_{1}, X_{2})\ge \frac{1}{2}\) and \(\mu _{R}(X_{2}, X_{3})\ge \frac{1}{2}\) then \( \mu _{R}(X_{1}, X_{3})\ge \frac{1}{2},\) where \(X_{1},X_{2}\) and \(X_{3}\) are \(2n+1\) fuzzy numbers.
From Lemmas 3.7 and 3.8, the extended fuzzy preference relation R is a total ordering relation (Epp (1990) and Lee (2005b)).
Lemma 3.9
Let \(X_{1}\) and \(X_{2}\) be two \(2n+1\) fuzzy numbers. By the extended fuzzy preference relation R, \(X_{1}\) is preferred to \(X_{2}\) if and only if \(\mu _{R}(X_{1}, X_{2})>\frac{1}{2}.\)
Lemma 3.10
\(X_{1}\succ X_{2}\) if and only if \(\mu _{R}(X_{1}, X_{2})>\frac{1}{2},\) where \(\succ \) is a binary relation.
Result 3.11
Let \(X_{1}=\big (a_{1}, a_{2},a_{3};\dfrac{1}{k}\big )\) and \(X_{2}=(b_{1}, b_{2},b_{3};\dfrac{1}{k}\big )\) be two generalized triangular fuzzy numbers. Then, \(\int _{0}^{\frac{1}{k}} [(X_{1}-X_{2})_{\alpha }^{L}+(X_{1}-X_{2})_{\alpha }^{U}]d\alpha =\dfrac{(a_{1}-b_{3})+2(a_{2}-b_{2})+(a_{3}-b_{1})}{2k}.\)
Now,
If \(k=1\) in the above, we get
Then, \(\mu _{R}(X_{1},X_{2}) =\frac{1}{2} ( \dfrac{(a_{1}-b_{3})+2(a_{2}-b_{2})+(a_{3}-b_{1})}{2\parallel T \parallel }+1)\) (Wang (2015b)).
Lemma 3.12
Let \(X_{1}=(a_{1}, a_{2},a_{3},a_{4},a_{5};\dfrac{2}{k})\) and \(X_{2}=(b_{1}, b_{2}, b_{3}, b_{4}, b_{5};\dfrac{2}{k})\) be two generalized pentagonal fuzzy numbers. Then, \(\mu _{R}(X_{1},X_{2}) = \frac{1}{2}\{[[(a_{1}-b_{5})+2(a_{2}-b_{4}) +2(a_{3}-b_{3}) +2(a_{4}-b_{2})+(a_{5}-b_{1})]/(2k\parallel T \parallel )]+1\}.\)
Proof
Now,
Then,
Similarly,
Then,
Now,
\(\square \)
Note 3.13
If \(k=2\) in Lemma 3.12, then \(\mu _{R}(X_{1},X_{2})\) \( = \frac{1}{2}\{[[(a_{1}-b_{5})+2(a_{2}-b_{4})+2(a_{3}-b_{3})+2(a_{4}-b_{2})+(a_{5}-b_{1})]/(4\parallel T \parallel )]+1\}.\)
Lemma 3.14
Let \(X_{1}=(a_{1}, a_{2},\ldots ,a_{n},a_{n+1},a_{n+2},\ldots ,a_{2n}\), \(a_{2n+1};\frac{2^{n-1}}{k})\) and \(X_{2}=(b_{1}, b_{2},\ldots ,b_{n},b_{n+1},\ldots ,b_{2n}\), \(b_{2n+1};\frac{2^{n-1}}{k})\) be two fuzzy numbers, where \(n\ge 3.\)
Then,
where
where \(t_{1}^{+}\)=max \(\{a_{1}, b_{1}\}\), \(t_{2}^{+}\)=max \(\{a_{2}, b_{2}\},\)...,\(t_{2n+1}^{+}\)=max\(\{a_{2n+1}, b_{2n+1}\},\) \(t_{1}^{-}\)=min \(\{a_{1}, b_{1}\}\), \(t_{2}^{-}\)=min \(\{a_{2}, b_{2}\},\)...,\(t_{2n+1}^{-}\)=min \(\{a_{2n+1}, b_{2n+1}\}.\)
Proof
Now, \(X_{1}-X_{2}=[(a_{1}-b_{2n+1}), (a_{2}-b_{2n}),\ldots ,(a_{n}-b_{n+2}), (a_{n+1}-b_{n+1}), (a_{n+2}-b_{n}),\ldots ,(a_{2n-1}-b_{3}),(a_{2n}-b_{2}),(a_{2n+1}-b_{1})].\)
Consider
Then,
and clearly, we get the value of \(\parallel T \parallel .\)
\(\square \)
3.5 Extension of relative preference relation on \(2n+1\) fuzzy numbers
Ranking n fuzzy numbers by fuzzy preference relation is time-consuming due to pair-wise comparisons. To reduce the time complexity, Wang (2015b) proposed relative preference relation. For example, to rank n fuzzy numbers by a preference relation, we require \(n_{C_{2}} \in O(n^{2})\) fuzzy pair-wise comparisons. In contrast, it is sufficient to use relative preference relation O(n) times to rank the fuzzy numbers. We extend the relative preference relation given by Wang (2015b) to rank \(2n+1\) fuzzy numbers by having the time complexity same as Wang’s method, i.e., the time complexity is O(n) for ranking n fuzzy numbers. The method is as follows:
Definition 3.16
Let \(A=\{X_{1},X_{2},\ldots ,X_{m}\}\) be m fuzzy numbers, where each \(X_{i}=\{x_{i1},x_{i2},\ldots ,x_{i(2n+1)}\},i=1,2,\ldots ,m\) and \({\overline{X}}=\frac{\sum _{i} X_{i}}{m}\) is the average of the m fuzzy numbers of A. An extended relative preference relation \(R_{*}\) with the fuzzy membership function \(\mu _{R_{*}}(X_{i}, {\overline{X}})\) represents the preference degree of \(X_{i}\) over \({\overline{X}}\) in A. Now, we define
where
where \(T^{+}_{q}\) is an interval \([t_{q1}^{+},t_{q2}^{+},\ldots ,t_{q(2n+1)}^{+}],\) \(T^{-}_{q}\) is an interval \([t_{q1}^{-}, t_{q2}^{-},\ldots ,t_{q(2n+1)}^{-}]\) and \(t_{q1}^{+}=\max \limits _{i}\{x_{i1}\},\)
If \(\mu _{R_{*}}(X_{i}, {\overline{X}}) > \frac{1}{2}\) then \(X_{i}\) is preferred to \({\overline{X}}\) and
\(\mu _{R_{*}}(X_{i}, {\overline{X}})< \frac{1}{2}\) then \({\overline{X}}\) is preferred to \(X_{i}.\)
Lemma 3.17
The extended relative preference relation \(R_{*}\) is a total ordering relation.
Lemma 3.18
Let \(X_{i}\) and \(X_{j}\) be two fuzzy numbers in A. Then, \(X_{i}\) is preferred to \(X_{j}\) if and only if \(\mu _{R_{*}}(X_{i}, {\overline{X}})>\mu _{R_{*}}(X_{j}, {\overline{X}}).\)
Lemma 3.19
\(X_{i}\succ X_{j}\) if and only if \(\mu _{R_{*}}(X_{i}, {\overline{X}})>\mu _{R_{*}}(X_{j}, {\overline{X}}),\) where \(\succ \) is a binary relation as defined in Definition 2.5.
Lemma 3.20
Let \(A=\{X_{1},X_{2},\ldots ,X_{m}\}\) be a set of fuzzy numbers, where each \(X_{i}=\{x_{i1},x_{i2},x_{i3},\ldots ,x_{i(2n+1)}\},i=1,2,\ldots ,m\) and \({\overline{X}}=({\overline{x}}_{1}, {\overline{x}}_{2},\ldots ,{\overline{x}}_{2n+1})\) be the average of m fuzzy numbers. The extended relative preference relation \(R_{*}\) with membership function \(\mu _{R_{*}}(X_{i}, {\overline{X}})\) represents preference degree of \(X_{i}\) over \({\overline{X}}\) in A. Then,
where
where
Here, we provide an example.
Example 3.21
Let \(X_{1}=(2,3,5,6,8;\frac{2}{3})\) and \(X_{2}=(1,2,4,7,9;\frac{2}{3})\) be two pentagonal fuzzy numbers (\(K=3\)).
Then,
Now,
Similarly,
Then,
This implies \(X_{1} > X_{2}\).
Now,
Now,
Similarly,
As \(\mu _{R_{*}}(X_{1},{\overline{X}})> \mu _{R_{*}}(X_{2},{\overline{X}}),\) we get \(X_{1}>X_{2}.\) We compared the above result with Wang and Lee (2008)’s area method.
By Wang and Lee (2008)’s method, we get \({\overline{x}}(X_{1})=4.75\) and \({\overline{x}}(X_{2})=4.6481.\) Therefore \(X_{1}>X_{2}.\)
4 Generalized 2n fuzzy number
4.1 Generalized linear 2n fuzzy number
Let \(\{a_{1}, a_{2},a_{3},\ldots ,a_{2n}\}\) be real numbers such that \(a_{1}<a_{2}<a_{3}<\cdots <a_{2n},~n=2,3,4,...\) and n is finite, \(k\ge 2^{n-2}.\) Now, we denote
Then, \([P(x),R_{2}(x),V_{2}(x)]\) gives fuzzy membership function of the generalized trapezoidal fuzzy number \((a_{1}, a_{2}, a_{3},a_{4};\dfrac{1}{k}).\) That is,
If \(k=1\) in the above, we get a trapezoidal fuzzy number.
Now,
gives fuzzy membership function of the generalized fuzzy number \((a_{1}, a_{2}, a_{3},\ldots ,a_{2n};\dfrac{2^{n-2}}{k}),\) where \(n\ge 3\) and \(k\ge 2^{n-2}.\)
In particular, if \(k=2^{n-2}\) then (2) gives fuzzy membership function of the fuzzy number \((a_{1}, a_{2}, a_{3},\ldots ,a_{2n};1),\) where \(n\ge 3.\)
For example, substitute \(n=3\) in (2) then \([P(x), S_{2}(x), R_{3}(x), U_{(3,1)}(x),V_{3}(x)]\) gives fuzzy membership function of the generalized hexagonal fuzzy number \((a_{1}, a_{2}, a_{3}, a_{4}, a_{5},a_{6};\dfrac{2}{k}).\)
That is,
If \(k=2\) in the above, we get a hexagonal fuzzy number (Fig. 2).
Similarly, substitute \(n=4\) in (2) then
gives fuzzy membership function of the generalized octagonal fuzzy number \((a_{1}, a_{2}, a_{3}, a_{4}, a_{5},a_{6},a_{7},a_{8};\dfrac{4}{k}).\)
That is,
If \(k=4\) in the above, we get an octagonal fuzzy number.
4.2 Generalized nonlinear 2n fuzzy number
Let \(\{a_{1}, a_{2},a_{3},\ldots ,a_{2n}\}\) be real numbers such that \(a_{1}<a_{2}<a_{3}<\cdots <a_{2n},~n=2,3,4,...\) and n is finite, \(m=1,2,...\) and m is finite. Now, take \(k\ge 2^{n-2}.\) Then, we denote
Now, \([P^{m}(x),R_{2}(x),V_{2}^{m}(x)]\) gives the fuzzy membership function of the generalized nonlinear trapezoidal fuzzy number \((a_{1}, a_{2}, a_{3},a_{4};\dfrac{1}{k}).\)
Now,
gives fuzzy membership function of the generalized nonlinear fuzzy number \((a_{1}, a_{2}, a_{3},\ldots ,a_{2n};\dfrac{2^{n-2}}{k}),\) where \(n\ge 3\) and \(k\ge 2^{n-2}.\)
In particular, if \(k=2^{n-2}\) then (4) gives fuzzy membership function of the nonlinear fuzzy number \((a_{1}, a_{2}, a_{3},\ldots ,a_{2n};1),\) where \(n\ge 3.\)
Note 4.3
If \(m=1\) in the fuzzy membership function of the generalized nonlinear 2n fuzzy number, then we get fuzzy membership function of the generalized linear 2n fuzzy number.
4.3 \(\alpha \)-cut of a generalized linear 2n fuzzy number
Let \(\{a_{1}, a_{2},a_{3},\ldots ,a_{2n}\}\) be real numbers such that \(a_{1}<a_{2}<a_{3}<\cdots <a_{2n},~n=1,2,3,...\) and n is finite. Now, take \(k\ge 2^{n-2}.\) We denote
Now, \([E(\alpha ), J_{1}(\alpha )]\) gives \(\alpha \)-cut of the generalized trapezoidal fuzzy number \((a_{1}, a_{2}, a_{3}, a_{4};\dfrac{1}{k}).\) That is,
If \(k=1\) in the above, we get an \(\alpha \)-cut of a trapezoidal fuzzy number \((a_{1}, a_{2}, a_{3}, a_{4};1)\).
Now,
gives \(\alpha \)-cut of the generalized fuzzy number \((a_{1}, a_{2}, a_{3},\ldots ,a_{2n};\dfrac{2^{n-2}}{k}),\) where \(n\ge 3\) and \(k\ge 2^{n-2}.\)
For example, if \(n=4\) in (6) then
gives \(\alpha \)-cut of the generalized octagonal fuzzy number \((a_{1}, a_{2}, a_{3}, a_{4}, a_{5},a_{6},a_{7},a_{8};\dfrac{4}{k}).\)
That is,
\(l(a_{1},a_{2},a_{3},a_{4}, a_{5},a_{6},a_{7},a_{8};\alpha )\)
If \(k=4\) (That is, \(k=2^{n-2}=2^{4-2}\)) in the above, we get an \(\alpha \)-cut of octagonal fuzzy number \((a_{1}, a_{2}, a_{3}, a_{4}, a_{5},a_{6},a_{7},a_{8};1).\)
Infimum and Supremum of \(\alpha \)-cut of a 2n fuzzy number:
Let \(B=(a_{1}, a_{2}, a_{3}, a_{4};\dfrac{1}{k})\) be a trapezoidal fuzzy number and \(\mu _{B}(x)\) is fuzzy membership function of B.
Then,
and
For a fuzzy number \((a_{1}, a_{2}, a_{3},\ldots ,a_{2n};\dfrac{2^{n-2}}{k}),\) \(B_{\alpha }^{L}=(E(\alpha ),F_{2}(\alpha ),F_{3}(\alpha ),\ldots ,F_{n-1}(\alpha ))\) and \(B_{\alpha }^{U}=(I_{(n,1)}(\alpha ), I_{(n+1,2)}(\alpha ), I_{(n+2,3)}(\alpha ),\ldots ,I_{(2n-3,n-2)}(\alpha ),J_{n}(\alpha )).\)
4.4 preference relation on 2n fuzzy numbers
We extend the fuzzy preference relation given by Wang (2015b) to rank 2n fuzzy numbers.
Definition 4.6
Let A and B be two 2n fuzzy numbers, where \(A=(a_{1},a_{2},\ldots ,a_{2n})\) and \(B=(b_{1},b_{2},\ldots ,b_{2n})\). An extended fuzzy preference relation \(R_{2n}\) is a subset of \({\mathbb {R}}\times {\mathbb {R}}\) with membership function \(\mu _{R_{2n}}(A,B)\) representing preference degree of A over B. Then,
where
\(T^{+}\) is an interval \([t_{1}^{+}, t_{2}^{+},\ldots ,t_{2n}^{+}],\) \(T^{-}\) is an interval \([t_{1}^{-}, t_{2}^{-},\ldots ,t_{2n}^{-}]\) and \(t_{1}^{+}=\max \{a_{1}, b_{1}\}\), \(t_{2}^{+}=\max \{a_{2}, b_{2}\}, \ldots ,t_{2n}^{+}=\max \{a_{2n}, b_{2n}\}\), \(t_{1}^{-}=\min \{a_{1}, b_{1}\}\), \(t_{2}^{-}=\min \{a_{2}, b_{2}\},\ldots ,t_{2n}^{-}=\min \{a_{2n}, b_{2n}\}.\)
In (II), if \(\int _{0}^{\frac{2^{n-2}}{k}}({(T^{+}-T^{-})^{\alpha }_{L}+(T^{+}-T^{-})^{\alpha }_{U}})d\alpha \ge 0,\) then \(\mu _{R_{2n}}(A, B)\ge \frac{1}{2}\) and if \(A=B,\) then \(\mu _{R_{2n}}(A, B)=\frac{1}{2}.\)
Lemma 4.7
The extended fuzzy preference relation \(R_{2n}\) is reciprocal. That is, \(\mu _{R_{2n}}(A, B)=1-\mu _{R_{2n}}(B, A)\) for all 2n fuzzy numbers A and B.
Lemma 4.8
The extended fuzzy preference relation \(R_{2n}\) is transitive. That is, if \(\mu _{R_{2n}}(A, B)\ge \frac{1}{2}\) and \(\mu _{R_{2n}}(B, C)\ge \frac{1}{2}\) then \( \mu _{R_{2n}}(A, C)\ge \frac{1}{2},\) where A, B and C are 2n fuzzy numbers.
From Lemmas 4.7 and 4.8, the extended fuzzy preference relation \(R_{2n}\) is a total ordering relation (Epp (1990) and Lee (2005b).
Lemma 4.9
Let A and B be two \(2n+1\) fuzzy numbers. By extended fuzzy preference relation \(R_{2n}\), A is preferred to B if and only if \(\mu _{R_{2n}}(A, B)>\frac{1}{2}.\)
Lemma 4.10
\(A\succ B\) if and only if \(\mu _{R_{2n}}(A, B)>\frac{1}{2},\) where \(\succ \) is a binary relation.
Lemma 4.11
Let \(A=(a_{1}, a_{2},a_{3},a_{4};\dfrac{1}{k})\) and \(B=(b_{1}, b_{2},b_{3},b_{4};\dfrac{1}{k})\) be two trapezoidal fuzzy numbers. Then, \(\int _{0}^{\frac{1}{k}} [(A-B)_{\alpha }^{L}+(A-B)_{\alpha }^{U}]d\alpha =\frac{1}{2k}[(a_{1}-b_{4}) +(a_{2}-b_{3})+(a_{3}-b_{2})+(a_{4}-b_{1})]\) and \(\mu _{R_{4}}(A,B)=\frac{1}{2}(\int _{0}^{\frac{1}{k}} (\dfrac{(A-B)^{\alpha }_{L}+(A-B)^{\alpha }_{U}}{\parallel T \parallel })d\alpha +1){=}\frac{1}{2} ( \dfrac{(a_{1}{-}b_{4}){+}(a_{2}{-}b_{3}){+}(a_{3}{-}b_{2}){+}(a_{4}-b_{1})}{2k\parallel T \parallel }{+}1).\) where
\(T^{+}\) is an interval \([t_{1}^{+}, t_{2}^{+},t_{3}^{+},t_{4}^{+}],\) \(T^{-}\) is an interval \([t_{1}^{-}, t_{2}^{-},t_{3}^{-},t_{4}^{-}]\) and \(t_{1}^{+}=\max \{a_{1}, b_{1}\}\), \(t_{2}^{+}=\max \{a_{2}, b_{2}\}\), \(t_{3}^{+}=\max \{a_{3}, b_{3}\}\), \(t_{4}^{+}=\max \{a_{4}, b_{4}\}\), \(t_{1}^{-}=\min \{a_{1}, b_{1}\}\), \(t_{2}^{-}=\min \{a_{2}, b_{2}\},t_{3}^{-}=\min \{a_{3}, b_{3}\}\) and \(t_{4}^{-}=\min \{a_{4}, b_{4}\}.\)
Lemma 4.12
Let \(A=(a_{1}, a_{2},a_{3},a_{4},a_{5},a_{6};\frac{2}{k})\) and \(B=(b_{1}, b_{2},b_{3},b_{4},b_{5},b_{6};\frac{2}{k})\) be two hexagonal fuzzy numbers. Then,
and
where
\(T^{+}\) is an interval \([t_{1}^{+}, t_{2}^{+},t_{3}^{+},t_{4}^{+},t_{5}^{+},t_{6}^{+}],\) \(T^{-}\) is an interval \([t_{1}^{-}, t_{2}^{-},t_{3}^{-},t_{4}^{-},t_{5}^{-},t_{6}^{-}]\) and \(t_{1}^{+}\)=max \(\{a_{1}, b_{1}\}\), \(t_{2}^{+}\)=max \(\{a_{2}, b_{2}\}\), \(t_{3}^{+}\)=max \(\{a_{3}, b_{3}\}\), \(t_{4}^{+}\)=max \(\{a_{4}, b_{4}\}\), \(t_{5}^{+}\)=max \(\{a_{5}, b_{5}\}\), \(t_{6}^{+}\)=max \(\{a_{6}, b_{6}\}\), \(t_{1}^{-}\)=min \(\{a_{1}, b_{1}\}\), \(t_{2}^{-}\)=min \(\{a_{2}, b_{2}\},t_{3}^{-}\)=min \(\{a_{3}, b_{3}\}\), \(t_{4}^{-}\)=min \(\{a_{4}, b_{4}\}\), \(t_{5}^{-}\)=min \(\{a_{5}, b_{5}\}\), \(t_{6}^{-}\)=min \(\{a_{6}, b_{6}\}.\)
Lemma 4.13
Let \(A=(a_{1}, a_{2},\ldots ,a_{n},a_{n+1},\ldots ,a_{2n-1}\), \(a_{2n};\frac{2^{n-2}}{k})\) and \(B=(b_{1}, b_{2},\ldots ,b_{n},b_{n+1},\ldots ,b_{2n-1},b_{2n};\) \(\frac{2^{n-2}}{k})\) be two fuzzy numbers, where \(n>3.\)
Then,
where
4.5 Extension of relative preference relation on 2n fuzzy numbers
Definition 4.15
Let \(A=\{X_{1},X_{2},\ldots ,X_{m}\}\) be a set of fuzzy numbers, where each \(X_{i}=\{x_{i1},x_{i2},x_{i3},\ldots ,x_{i(2n)}\},i=1,2,\ldots ,m.\) and \({\overline{X}}=({\overline{x}}_{1}, {\overline{x}}_{2},\ldots ,{\overline{x}}_{2n})\) be the average of m fuzzy numbers. The relative preference relation \(R_{2n}^{*}\) with membership function \(\mu _{R_{2n}^{*}}(X_{i}, {\overline{X}})\) represents preference degree of \(X_{i}\) over \({\overline{X}}\) in A.
Then,
where
Example 4.16
Let
be octagonal fuzzy numbers.
Then, \(\parallel T \parallel \) of \(X_{1}\) and \(X_{2}\) is
where
Then,
Therefore \(X_{1}\) is preferred to \(X_{2}.\)
Similarly, \( \mu _{R_{8}}(X_{1},X_{3})=0.4924<\frac{1}{2}\).
This implies \(X_{3}\) is preferred to \(X_{1}\) and \(\mu _{R_{8}}(X_{2},X_{3})=0.4885<\frac{1}{2}.\) This implies \(X_{3}\) is preferred to \(X_{2}.\) Therefore \(X_{3} \succ X_{1} \succ X_{2}.\)
By extended relative preference relation,
Similarly,
and
Now,
Similarly, we get \(\mu _{R_{8}}^{*}(X_{2}, {\overline{X}})=0.4939\) and \(\mu _{R_{8}}^{*}(X_{3}, {\overline{X}})=0.5070.\)
As \(\mu _{R_{8}}^{*}(X_{3}, {\overline{X}})>\mu _{R_{8}}^{*}(X_{1}, {\overline{X}})>\mu _{R_{8}}^{*}(X_{2}, {\overline{X}}),\) we get \(X_{3}\succ X_{1} \succ X_{2}.\)
The above result is compared with Wang and Lee (2008)’s method and according to this method the values are \({\overline{x}}(X_{1})=0.2954\), \({\overline{x}}(X_{2})=0.2853\) and \({\overline{x}}(X_{3})=0.2976.\)
Hence we get \(X_{3}\succ X_{1} \succ X_{2}.\)
5 Fuzzy multi-criteria decision-making (FMCDM) model
In this section, we extend FMCDM model given by Wang (2014) using the relative preference relation on \(2n+1\) fuzzy numbers.
In this algorithm, we consider the generalized \(2n+1\) fuzzy numbers with \(k=2^{n-1}\) and we denote them as \((a_{1}, a_{2},\ldots ,a_{2n+1})\) instead of \((a_{1}, a_{2},\ldots ,a_{2n+1};1).\)
We take \(E_{1},E_{2},\ldots ,E_{r}\) as the experts who provide their opinion on criteria \(C_{1}, C_{2},\ldots ,C_{t}\) of the alternatives \(A_{1},A_{2},\ldots ,A_{p}\). Let \(B_{ijl}{=}(b_{ijl1},b_{ijl2},b_{ijl3},\ldots ,b_{ijl(2n{+}1)})\) be the evaluation rating given by the expert \(E_{l}\) for alternative \(A_{i}\) on criterion \(C_{j},\) where \(i=1,2,\ldots ,p\), \(j=1,2,\ldots ,t\), \(l=1,2,\ldots ,r.\)
Then,
where
The normalized value of \(B_{ij}\) is denoted by \({\widetilde{B}}_{ij}\) and it is classified as follows.
If \(B_{ij}\) belongs to cost criteria then
where
If \(B_{ij}\) belongs to benefit criteria then
where
Let \(W_{jl}=(w_{jl1}, w_{jl2},\ldots ,w_{jl(2n+1)})\) be the weight of the criterion \(C_{j}\) given by the expert \(E_{l},\) where
Then,
where
The normalized values of \(i^{th}\) alternative on t criteria are presented as follows.
Then, we find ideal and anti-ideal solutions according to as Wang et al. (2003).
The ideal solution is \(A^{+}=[{\widetilde{B}}_{1}^{+}, {\widetilde{B}}_{2}^{+},\ldots ,{\widetilde{B}}_{t}^{+}],\) where \({\widetilde{B}}_{j}^{+}=(t_{qj1}^{+}, t_{qj2}^{+},\ldots ,t_{qj(2n+1)}^{+})\) is the best rating on jth criterion for all normalized values, \(j=1,2,\ldots ,t.\)
The anti-ideal solution is \(A^{-}=[{\widetilde{B}}_{1}^{-}, {\widetilde{B}}_{2}^{-},\ldots ,{\widetilde{B}}_{t}^{-}],\) where \(B_{j}^{-}=(t_{qj1}^{-},t_{qj2}^{-},\ldots ,t_{qj(2n+1)}^{-})\) is the worst rating on jth criterion for all normalized values, \(j=1,2,\ldots ,t.\)
Now, \(\mu _{R_{*}}({\widetilde{B}}_{j}^{+},{\widetilde{B}}_{ij})\) indicates the relative preference degree of \(A^{+}\) over the alternative \(A_{i}\) to the jth criterion,
Then,
and
indicates relative preference degree of alternative \(A_{i}\) over \(A^{-}\) to the jth criterion, where
Table 4 represents linguistic terms of weights of criteria given by experts Tables 5, 6 and 7 provide related calculations.
where
and
\(D_{i}^{+}\) is derived by using \(\mu _{R_{*}}({\widetilde{B}}_{j}^{+}, {\widetilde{B}}_{ij})\) and \(W_{j},~j=1,2,\ldots ,t\) and indicates the weighted preference degree of \(A^{+}\) over the alternative \(A_{i}\) and \(D_{i}^{-}\) is derived by using \(\mu _{R_{*}}({\widetilde{B}}_{ij},{\widetilde{B}} _{j}^{-})\) and \(W_{j},~j=1,2,\ldots ,t\) and indicates the weighted preference degree of \(A_{i}\) over \(A^{-}\) for \(i=1,2,\ldots ,p.\)
Now,
and \(D_{i}^{-}=(W_{1}\mu _{R_{*}}({\widetilde{B}}_{i1}, {\widetilde{B}}_{1}^{-}))+(W_{2}\mu _{R_{*}}({\widetilde{B}}_{i2}, {\widetilde{B}}_{2}^{-}))+\cdots +(W_{t}\mu _{R_{*}}({\widetilde{B}}_{it}, {\widetilde{B}}_{t}^{-})),~~i=1,2,\ldots ,p.\)
Then, we get \(D_{i}^{+}\) and \(D_{i}^{-}\) as \(2n+1\) fuzzy numbers and are denoted by
Then,
where
and
Now, the relative closeness coefficient \(D_{i}\) of the alternative \(A_{i}\) is defined as
where \(\mu _{R_{*}}(D_{i}^{-}, {\overline{D}}^{-})\) is the relative preference degree of \(D_{i}^{-}\) over \({\overline{D}}^{-}\) and \(\mu _{R_{*}}(D_{i}^{+}, {\overline{D}}^{+})\) is the relative preference degree of \(D_{i}^{+}\) over \(D^{+}.\)
Clearly, \(D_{i},~i=1,2,\ldots ,p\) is in the interval [0, 1]. The bigger the value \(D_{i},~i=1,2,\ldots ,p\) is closer the ideal solution is. Alternatively, smaller the value \(D_{i},~i=1,2,\ldots ,p\) is closer the anti-ideal solution is. Thus, the P alternatives are ranked according to their relative closeness coefficients \(D_{1}, D_{2},\ldots ,D_{p}.\)
Analysis of computational complexity
The computational complexity of the proposed algorithm is O(mn), where m is the number of alternatives with n criteria. Like in the other FMCDM methods, it is assumed that the number of experts is a constant for the purpose of computational complexity. If \(m=O(n)\), then the algorithm takes quadratic \((O(n^{2}))\) time. In general, FMCDM methods have O(mn) computational complexity. This coincides with the complexity of the algorithm given in this paper.
Example 5.1
In real-world conditions, selection of an internet service provider is often based on various criteria such as Monthly cost (\(C_{5}\)) as Cost criteria and Volume of data (\(C_{1}\)), Speed of internet (\(C_{2}\)), Subscription to OTT platforms (\(C_{3}\)), dependable customer service (\(C_{4}\)) as benefit criteria. If the available internet service providers (\(A_{1}\), \(A_{2}\), \(A_{3}\), \(A_{4}\)) are assessed on this criterion by four different experts (\(E_{1}\), \(E_{2}\), \(E_{3}\), \(E_{4}\)) with each expert considering one of the benefit criteria to be more important than others, then there will be a dilemma in the selection as each service provider may have at least one parameter in which they are superior to other internet service providers. In such situations, we provide a solution to arrive at a decision as follows.
Table 1 represents the consolidated information about the performance of various internet service providers (\(A_{1}\), \(A_{2}\), \(A_{3}\), \(A_{4}\)) with respect to various selection criteria (\(C_{1}\), \(C_{2}\), \(C_{3}\), \(C_{4}\), \(C_{5}\)) by various experts (\(E_{1}\), \(E_{2}\), \(E_{3}\), \(E_{4}\)) wherein the feedback on benefit criteria is expressed as Very Good (VG), Good (G), Average (A), Poor (P) or Very Poor (VP). The feedback on cost criteria, on the other hand, is expressed as numerical values (for example, here it represents price of the service).
Linguistic terms used for expressing the feedback on benefit criteria are assigned with fuzzy numbers as shown in Table 2.
Table 3 represents weights of criteria and their corresponding fuzzy numbers.
Ideal and anti-ideal solutions of 4 internet service providers are presented in Table 8. Subsequent calculations are shown in Tables 9, 10, 11, 12, 13, 14, 15, 16.
From Table 17, the relative closeness coefficients of 4 internet service providers are \(A_{1}: 0.4993\), \(A_{2}: 0.5767,\) \(A_{3}: 0.3607\) and \(A_{4}: 0.5624\) and hence the rank order of internet service providers is \(A_{2}>A_{4}>A_{1}>A_{3}.\) This indicates that the internet service provider \(A_{2}\) is better compared to other internet service providers.
6 Validation with existing methods
Wang (2014) proposed an algorithm to give decision in a triangular FMCDM using relative preference relation along with TOPSIS method. In the present study, we extended Wang’s method to arrive at a decision in FMCDM problems involving \(2n +1\) fuzzy numbers. Similar extension is possible for FMCDM problems involving 2n fuzzy numbers.
The suitability of the proposed method was verified by comparing it with the popularly used multi-criteria decision-making methods such as VIKOR, MOORA and ELECTRE. Prior to the application of these methods, alternative values of various criteria given by experts and weights of criteria from the above example are converted to crisp values by centroid defuzzification method.
1.VIKOR: Following table of values is obtained by applying VIKOR method (Opricovic (1998, 2002)).
\(A_{1}\) | \(A_{2}\) | \(A_{3}\) | \(A_{4}\) | |
---|---|---|---|---|
S | 1.5384 | 0.3488 | 2.3193 | 0.5928 |
R | 0.6950 | 0.2182 | 0.6950 | 0.4504 |
Q | 0.8019 | 0 | 1 | 0.3054 |
From the above table, rank order of the alternatives is \(A_{2}>A_{4}>A_{1}>A_{3}.\)
2. Multi-objective Optimization by Ratio analysis (MOORA): By applying MOORA reference point method ( Brauers and Zavadskas (2006, 2010)), deviations from the reference points and ranking of the alternatives are presented in the following table.
\(C_{1}\) | \(C_{2}\) | \(C_{3}\) | \(C_{4}\) | \(C_{5}\) | max. value | Ranking | |
---|---|---|---|---|---|---|---|
\(A_{1}\) | 0 | 0.0386 | 0.11 | 0.0682 | 0 | 0.11 | 3 |
\(A_{2}\) | 0.0527 | 0 | 0 | 0 | 0.0404 | 0.0527 | 1 |
\(A_{3}\) | 0.1329 | 0.0793 | 0.0535 | 0.0211 | 0.1111 | 0.1329 | 4 |
\(A_{4}\) | 0.0775 | 0.0203 | 0.0253 | 0.0235 | 0.005 | 0.0775 | 2 |
3. ELECTRE: By using ELECTRE method (Roy (1991)), following global matrix is obtained.
![](http://media.springernature.com/lw188/springer-static/image/art%3A10.1007%2Fs00500-021-06616-1/MediaObjects/500_2021_6616_Equ101_HTML.png)
From the global matrix, it is clear that \(A_{2}>A_{3}>A_{1}\) and \(A_{4}>A_{3}>A_{1}\) which can also be expressed as \(A_{2}\ge A_{4}>A_{3}>A_{1}\).
The rank order of alternatives obtained from all three methods indicates the proposed method’s suitability in FMCDM problems. The proposed method involves less operational complexity, it is easy to compute, and it minimizes the loss of information.
7 Conclusions
We have defined \(2n+1\) and 2n fuzzy numbers as generalizations of triangular and trapezoidal fuzzy numbers, respectively. Fuzzy preference relation and relative preference relation of Wang (2015b) are extended to rank 2n and \(2n+1\) fuzzy numbers and the results are compared with Wang and Lee (2008) method. Wang (2014)’s method was extended to arrive at a decision in FMCDM problems when the given data is in terms of \(2n+1\) fuzzy numbers. An illustrative example was provided to explain the suitability of the proposed method, and the results were validated using VIKOR, MOORA and ELECTRE methods. The proposed method can also be extended to 2n fuzzy numbers.
References
Asady B, Zendehnam A (2007) Ranking fuzzy numbers by distance minimization. Appl Math Model 31:2589–2598
Brauers WKM, Zavadskas EK (2006) The MOORA method and its application to privatization in a transition economy. Control Cybern 35(2):445–469
Brauers WKM, Zavadskas EK (2010) Project management by multimoora as an instrument for transition economies. Technol Econ Dev Econ 16(1):5–24. https://doi.org/10.3846/tede.2010.01
Chen YC, Huang WC (2021) Constructing a stock-price forecast CNN model with gold and crude oil indicators. Appl Soft Comput 112:107760
Cheng CH (1998) A new approach for ranking fuzzy numbers by distance method. Fuzzy Sets Syst 95:307–317
Choobineh F, Li H (1993) An index for ordering fuzzy numbers. Fuzzy Sets Syst 54:287–294
Chu TC, Tsao CT (2002) Ranking fuzzy numbers with an area between the centroid point and the original point. Comput Math Appl 43:111–117
Dias OP Jr (1993) Ranking alternatives using fuzzy numbers: a computational approach. Fuzzy Sets Syst 56:247–252
Epp HS (1990) Discrete Mathematics with Applications. Wadsworth, California
Fortemps P, Roubens M (1996) Ranking and defuzzification methods based on area compensation. Fuzzy Sets Syst 82:319–330
Jain R (1976) Decision-making in the presence of fuzzy variables. IEEE Trans Syst Man Cybern 6:698–703
Jain R (1977) A procedure for multiple-aspect decision making using fuzzy sets. Int J Syst Sci 8(1):1–7
Koppula K, Kedukodi BS, Kuncham SP (2019) Markov chains and rough sets. Soft Comput 23(15):6441–6453
Koppula K, Kedukodi BS, Kuncham SP (2020) Markov frameworks and stock market decision making. Soft Comput 24:16413–16424
Lalotra S, Singh S (2020) Knowledge measure of hesitant fuzzy set and its application in multi-attribute decision-making. Comput Appl Math 39:86
Lee ES, Li RJ (1988) Comparison of fuzzy numbers based on the probability measure of fuzzy events. Comput Math Appl 15:887–896
Lee HS (2005) A fuzzy multi-criteria decision making model for the selection of the distribution center. Lect Notes Comput Sci 3612:1290–1299
Lee HS (2005) On fuzzy preference relation in group decision making. Int J Comput Math 82:133–140
Mandal P, Ranadive AS (2019) Fuzzy multi-granulation decision-theoretic rough sets based on fuzzy preference relation. Soft Comput 23:85–99
Opricovic S (1998) Multicriteria Optimization of Civil Engineering Systems. Faculty of Civil Engineering, Belgrade 2:5–21
Opricovic S (2002) Multicriteria planning of post-earthquake sustainable reconstruction. Comput Aided Civ Infrastruct Eng 17:211–220
Riaz M, Davvaz B, Fakhar A, Atiqa F (2020) Hesitant fuzzy soft topology and its applications to multi-attribute group decision-making. Soft Comput 24:16269–16289
Roy B (1991) The outranking approach and thinks of ELECTRE methods. Theor Decis 31:49–73
Singh S, Lalotra S, Sharma S (2019) Dual concepts in fuzzy theory: Entropy and knowledge measure. Int J Intell Syst 34(5):1034–1059
Singh S, Sharma S (2019) On generalized fuzzy entropy and fuzzy divergence measure with applications. Int J Fuzzy Syst Appl 8(3):47–69
Singh S, Sharma S, Ganie AH (2020) On generalized knowledge measure and generalized accuracy measure with applications to MADM and pattern recognition. Comput Appl Math 39:231
Wang YJ, Lee HS, Lin K (2003) Fuzzy TOPSIS for multi-criteria decision-making. Int Math J 3:367–379
Wang YJ, Lee HS (2008) The revised method of ranking fuzzy numbers with an area between the centroid and original points. Comput Math Appl 55:2033–2042
Wang YJ (2014) A fuzzy multi-criteria decision-making model by associating technique for order preference by similarity to ideal solution with relative preference relation. Inf Sci 268:169–184
Wang YJ (2015) A fuzzy multi-criteria decision-making model based on simple additive weighting method and relative preference relation. Appl Soft Comput 30:412–420
Wang YJ (2015) Ranking triangle and trapezoidal fuzzy numbers based on the relative preference relation. Appl Math Model 39:586–599
Wang YJ (2020) Combining technique for order preference by similarity to ideal solution with relative preference relation for interval-valued fuzzy multi-criteria decision-making. Soft Comput 24:11347–11364
Wang YJ (2020) Utilization of trapezoidal intuitionistic fuzzy numbers and extended fuzzy preference relation for multi-criteria group decision-making based on individual differentiation of decision-makers. Soft Comput 24:397–407
Zadeh LA (1965) Fuzzy sets. Inf Control 8:338–353
Zimmermann HJ (1987) Fuzzy sets. Decision making and Expert Systems. Kluwer, Boston, p 1987
Zimmermann HJ (1991) Fuzzy set theory and its applications, 2nd edn. Kluwer, Boston, p 1991
Acknowledgements
The authors thank the anonymous reviewers for their valuable comments and suggestions. All authors acknowledge the support and encouragement of Manipal Institute of Technology, Manipal Academy of Higher Education (MAHE), Manipal.
Funding
Open access funding provided by Manipal Academy of Higher Education, Manipal.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflicts of interest
The authors declare that they have no conflict of interest.
Ethical approval
This article does not contain any studies with human participants or animals performed by any of the authors.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Koppula, K., Kedukodi, B.S. & Kuncham, S.P. Generalization and ranking of fuzzy numbers by relative preference relation. Soft Comput 26, 1101–1122 (2022). https://doi.org/10.1007/s00500-021-06616-1
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00500-021-06616-1