You are on page 1of 23

Int. J. Fuzzy Syst.

https://doi.org/10.1007/s40815-023-01652-5

Prospect Theory-Based q-Rung Orthopair Fuzzy TODIM Method


for Risk Assessment of Renewable Energy Projects
Yi Liu1,2 • Ya Qin1 • Haobin Liu1,2 • Saleem Abdullah3 • Yuan Rong4

Received: 26 July 2023 / Revised: 14 October 2023 / Accepted: 14 November 2023


 The Author(s) under exclusive licence to Taiwan Fuzzy Systems Association 2024

Abstract Traditional fossil fuels and has become an q-ROF Lance distance measure is propounded for con-
important component of clean energy. Vigorously invest- quering the deficiencies caused by the prior distances.
ing in and developing renewable energy projects is of great Then, an improved weight model by the defined q-ROF
strategic significance for enhancing sustainable energy Lance distance is applied for determining the weight of
development. Due to the unique nature of energy project experts, and an integration model that incorporates the
investment, evaluating the investment risks of renewable group individual best and worst (GIBWM) and criteria
energy projects is crucial for enhancing project feasibility. importance through inter-criteria correlation (CRITIC) is
Therefore, this study propounds a novel multiple-criteria presented for ascertaining the synthetics weight of criteria.
group decision framework based on the extended TODIM Afterward, an extended TODIM method based upon the
method (an acronym in Portuguese for Interactive and prospect theory using the proposed Lance distance measure
Multi-criteria Decision Making) under q-rung orthopair is developed to rank alternatives. Furthermore, a hybrid
fuzzy (q-ROF) set for the application of renewable energy decision framework is constructed to assess the investment
project investment risk assessment. To begin with, a novel risk of renewable energy projects. The sensitivity analysis
regarding the risk avoidance coefficient is conducted to
explore the stability and robustness of the propounded
& Yuan Rong assessment framework. Further, the extant methodologies
rongyuanry@163.com are utilized to compare with the designed decision frame-
Yi Liu work to highlight its strengths and superiorities. The out-
liuyiyl@126.com comes exhibit that the propounded framework can provide
Ya Qin efficient decision support for assessing renewable energy
qinyaqy@126.com project investment risk by modeling the uncertainty and
Haobin Liu psychological behavior of the assessment procedure.
njtcliuhb@163.com
Saleem Abdullah Keywords q-ROF set  MCGDM  GIBWM  CRITIC 
saleemabdullah@awkum.edu.pk Extended TODIM
1
School of Mathematics and Information Sciences, Neijiang
Normal University, Neijiang 641000, China
2
1 Introduction
Numerical Simulation Key Laboratory of Sichuan Provincial
Universities, Data Recovery Key Laboratory of Sichuan
Province, Neijiang Normal University, Neijiang 641000, With the rapid development of the global economy and the
China increasing shortage of traditional fuel energy caused by
3
Department of Mathematics, Abdul Wali Khan University, environmental and ecological issues, the pressure on
Mardan 23200, KP, Pakistan energy conservation and emission reduction in different
4
School of Management, Shanghai University, energy industries is increasing. Therefore, vigorously pro-
Shanghai 200444, China moting the implementation and development of renewable

123
International Journal of Fuzzy Systems

energy projects is an important measure for the country to problems and laid a theoretical foundation for the study of
implement the policy of energy conservation and emission fuzzy decision. Afterward, to characterize the ambiguity in
reduction. Renewable energy involves projects such as practical problems from multiple perspectives and improve
solar power generation, hydroelectric power generation, the accuracy of decision information, Atanassov [12] pro-
and biomass energy generation. Due to its low pollution, posed intuitionistic fuzzy set (IFS) theory by extending
renewable energy, energy conservation, emission reduction traditional fuzzy sets to more finely express fuzziness.
advantages, and enormous development potential, it has Although IFS models the vagueness from three aspects of
gradually attracted widespread attention from many membership, non-membership, and hesitancy simultane-
investment enterprises in recent years. Due to a large ously, it also has certain limitations, as the sum of mem-
amount of investment in energy projects and their close bership and non-membership is limited to less than or equal
relationship with social harmony and development, to one. To conquer the above-mentioned shortcomings, the
Therefore, conducting an effective comprehensive evalua- extensions of IFS including inter-valued IFS [13], Pytha-
tion of investment risks and selecting the most suit- gorean fuzzy set (PFS) [14] including inter-valued PFS
able project is an important issue that enterprises urgently [15] and q-ROF set [16] are propounded to expand the
need to pay attention to. Due to the need to comprehen- information expression range and provide more options for
sively consider the preferences of multiple experts for experts to provide their preferences. From these extensions,
alternative projects under different dimensional criteria in the q-ROF set possesses powerful flexibility and is uni-
the evaluation of investment risks, as well as the com- versal to depict uncertain information and has been
plexity and uncertainty of conflicting criteria and evalua- investigated to achieve numerous valuable outcomes. In
tion environment, investment risk assessment of renewable terms of fundamental theory of q-ROF set, the aggregation
energy projects is regarded as a fuzzy multi-criteria group operators [17, 18], similarity measure [19], entropy mea-
decision-making problem and has received significant sure [20], decision methodology [21], and applications of
attention from scholars [1–5]. For instance, Hashemizadeh q-ROF set [22, 23] has been studies sequentially. The
et al. [6] constructed the risk assessment framework based distance measure is also an essential factor to measure the
on fuzzy-analytic network process, COPRAS, MABAC, differences between q-ROF sets. Du [24] proposed the
and GRA methods to determine the ranking of renewable Minkowski-type distance measures for the q-ROF set and
energy sources under uncertain environment. Wu et al. [7] explored its continuous universe and the bounded and
introduced a fused hesitant fuzzy linguistic term set cloud unbounded case examples. Peng and Liu [25] introduce a
model approach for the risk assessment of renewable series of distance measures, and other information mea-
energy projects by considering the vagueness and ran- sures and applied them to medical diagnosis, clustering
domness of evaluation information. Karamoozian et al. [8] analysis, and so forth. Afterward, Liu et al. [19] defined
propounded a new hybrid failure mode and effect analysis some cosine similarity measures and distance measures for
approach to evaluate the investment risk of renewable q-ROF set based on the cosine function. Liao et al. [26]
energy projects. Li et al. [9] evaluated the renewable propounded an innovative q-ROF distance measure and
energy investment risk based on the system dynamics and constructed a novel q-ROF GLDS group decision method
fuzzy analytic hierarchy process approach under uncertain for its application in investment evaluation. Zeng et al. [27]
circumstances. Karamoozian et al. [10] reported a hybrid introduced a q-ROF weighted induced logarithmic distance
approach by merging DEMATEL and QUALIFLEX measure and built a novel MCDM method for the selection
methods within interval type 2 hesitant fuzzy information, of smart phone. Given the mentioned analysis, It is obvious
which offers a valuable model for enterprises and man- that distance measures in the q-ROF set has been exten-
agers. The aforesaid approaches and models serve as effi- sively studied and applied to decision approach construc-
cient decision support for investors to appraise the tion and cluster analysis. However, few research explored
investment risk of renewable energy projects. However, the the extension and application of the Lance distance mea-
mentioned studies fail to think about the psychological sure between q-ROF sets. Hence, it is necessary to incor-
behavior of investors and the combination criteria weight porate the Lance distance into q-ROF set for providing an
model within an uncertainty setting. Hence, this study will advantageous distance to unfold decision analysis.
try to fill these gaps and design a powerful assessment To resolve MCDM issues, numerous classical decision
framework to conquer the found shortcomings. models including the weighted sum model, TOPSIS and
The conflict and uncertainty of actual decision problems VIKOR are often utilized to employ in various fields for
pose certain challenges and constraints for decision-makers unfold decision analysis. Due to the powerful merit of the
when analyzing actual decision-making problems. Based q-ROF set in treating uncertainty, numerous scholars
on this inspiration, the founding of fuzzy set (FS) theory extended the traditional decision methodology to the
[11] effectively characterized uncertainty in practical q-ROF setting and received widespread applications.

123
Y. Liu et al.: Prospect Theory-Based q-Rung...

Krishankumar et al. [28] reported a novel renewable energy powerful performance for unfolding uncertain decision
sources assessment framework based on the COPRAS analysis. Nevertheless, there is no investigation to
method and q-ROF set. Darko and Liang [29] proposed an combine the extended TODIM approach and prospect
extended EDAS method based on some novel Hamacher theory within q-ROF set.
operators for establishing MCGDM under the q-ROF set- • Prior studies have introduced several approaches for
ting. Rani and Mishra [30] presented an expanded WAS- evaluating the renewable energy project investment
PAS method within q-ROF set based on novel score risks under diverse uncertain environments but these
function and similarity for choosing the optimal fuel methods have some defects in settling the uncertain and
technology. Ali [31] developed a novel MCGDM method psychological preference during the evaluation
by merging MARCOS and CRITIC methods based on a procedure.
novel score function for solid waste management. How- With the aid of the discussed motivations, the main con-
ever, the mentioned approaches assume that experts are tributions and aims of this research are outlined as follows:
completely rational during the course of a decision. In
order to the psychological behavior of experts in decision • A novel q-ROF Lance distance measure is propounded
analysis, Gomes [32] propounded the TODIM method to distinguish the difference between two q-ROF sets
based on prospect theory to consider the psychological and its merits than the extant works are also discussed;
behavior of experts and effectively solve decision prob- • A novel model based on the defined q-ROF Lance
lems. Since then, TODIM approach is extended to diverse distance measure for estimating the weight coefficient
uncertain circumstances, such as PFS [33], IFS [34], of experts is presented and a combination weight model
q-ROF [35], probabilistic uncertain linguistic [36], proba- that incorporates the q-ROF-GIBWM and q-ROF-
bilistic hesitant fuzzy set [37, 38], Z-mixture numbers [39]. CRITIC methods to ascertain the ultimate weight
Meanwhile, the TODIM method is also applied to various coefficient of criteria;
practical applications including portfolio allocation [40], • An extended TODIM method based upon the prospect
barriers assessment [41], healthcare service quality man- theory using the proposed Lance distance measure is
agement [42], and wind turbine system risk ranking [43]. developed for attaining the ranking of considered schemes;
The TODIM method has been applied to uncertain decision • A hybrid decision framework called q-ROF-GIBWM-
analysis successfully. However, the value function in the CRITIC-TODIM is constructed to unfold uncertain
classical TODIM method cannot better express the loss decision analysis;
aversion behavior of experts. Hence, this study utilizes the • A case that assesses the investment risk of renewable
logarithm value function to compute the dominance of energy projects is implemented to validate the feasi-
schemes and further proposed the extended TODIM bility and practicability of the proposed hybrid q-ROF-
approach. Based on the mentioned analysis, it is noticed GIBWM-CRITIC-TODIM decision framework.
that the TODIM method has not yet been combined with The remainder of this article is expounded as follows:
prospect theory to evaluate the investment risk for renew- Section 2 succinctly retrospects some fundamental con-
able energy projects. Therefore, it is an important cepts of the q-ROF set. Section 3 propounds a novel q-ROF
advancement to propose an extended TODIM method Lance distance measure and analyses its strengths of it and
based on prospect theory by using the Lance distance the extant distance measures. Section 4 establishes a hybrid
measure under q-ROF circumstances. q-ROF-GIBWM-CRITIC-TODIM decision framework and
Using the mentioned discussions and investigations, the illustrates its detailed procedures. In Sect. 5, the developed
motivations of this study are displayed as follows: framework is employed to evaluate the renewable energy
• The previous q-ROF distance measures are invalid in project investment risk to show its applicability, contrastive
some special situations and affect the reasonability of and sensitivity studies are also carried out Lastly, several
the ultimate decision outcomes. Hence, a novel distance conclusions, limitations, and prospects are depicted in Sect.
should be introduced to conquer the aforesaid defects. 6.
Besides, there is no research about Lance distance
measures under q-ROF circumstances.
• In the literature, no research devotes to establishing the 2 Preliminaries
combined criteria weight-determination model via
incorporating the GIBWM and CRITIC method in Some basic concepts of the q-ROF set will be reviewed in
q-ROF setting. this part, which is the basis of the present work.
• The extended TODIM approach is an enhanced work Definition 1 [17] Let X be a finite universe of discourse.
by generalizing the TODIM approach and possesses A q-ROF set A can be expressed as

123
International Journal of Fuzzy Systems

A ¼ fðl= ðxi Þ; m= ðxi ÞÞjx 2 X g ð1Þ q-ROFWGð=1 ; =2 ; . . .; =n Þ


0 vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1
Y
L   u Y n 
where l= ðxi Þ; m= ðxi Þ 2 ½0; 1 are the membership grade and - u  q - j ð5Þ
¼ @ lj ; t1 
j q
1  mj A:
non-membership grade of x in q-ROF set =, respectively,
j¼1 j¼1
with 0  ðl= ðxi ÞÞq þðm= ðxi ÞÞq  1. The hesitancy grade
1
p= ðxi Þ is defined by p= ðxi Þ ¼ ð1  ðl= ðxi ÞÞq ðm= ðxi ÞÞq Þq .
The pair ðl= ðxi Þ; m= ðxi ÞÞ is denoted as a q-ROF number.
For convenience, = ¼ ðl= ; m= Þ denotes a q-ROF
number[17]. 3 Novel Distance Measure for q-ROF Information
Definition 2 [17] For two arbitrary q-ROF numbers =1 ¼
This section first reviews some extant q-ROF distance
ðl1 ; m1 Þ and =2 ¼ ðl2 ; m2 Þ, the basic operational laws are
measures and analyses the deficiencies of these works.
defined as below:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi  Then a novel distance measure called q-ROF Lance dis-
q q q tance is propounded to conquer the mentioned defects.
(i) =1  =2 ¼ 1  ð 1  ð l1 Þ Þ ð 1  ð l 2 Þ Þ ; m 1 m 2 ;
 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 3.1 Analyses of Existing Distances for q-ROF
(ii) =1  =2 ¼ l1 l2 ; 1  ð1  ðm1 Þq Þð1  ðm2 Þq Þ ;
q
Information
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 
q q j j
(iii) j=1 ¼ 1  ð1  ðl1 Þ Þ ; ðm1 Þ ; j [ 0; Distance measures are usually employed to describe the
 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi difference between two q-ROF sets. Many distance mea-
j q j
(iv) =1 ¼ ðl1 Þ ; 1  ð1  ðm1 Þq Þ ; j [ 0;
j sures are proposed in the q-ROF environment and applied
to related fields, such as pattern recognition, decision-
(v) =c1 ¼ ðm1 ; l1 Þ: making. In what follows, some popularly used distances are
listed below.
It is supposed that X ¼ fx1 ; . . .; xn g be a district finite
Definition 3 [17] Given = ¼ ðl; mÞ is a q-ROF number, universe of discourse. A ¼ fðlA ðxi Þ; mA ðxi ÞÞjxi 2 X g, B ¼
the score SF ð=Þ and accuracy functions AF ð=Þ are defined fðlB ðxi Þ; mB ðxi ÞÞjxi 2 X g are two q-ROF sets on X. Then
as the existing q-ROF distance measures are reviewed as
follows:
SF ð=Þ ¼ lq  mq ; SF ð=Þ 2 ½1; 1; ð2Þ
(1) Minkowski distance of q-ROF sets presented by Du
AF ð=Þ ¼ lq þ mq ; AF ð=Þ 2 ½0; 1: ð3Þ
[24].
!1p
1 X 1 X
dDu ðA; BÞ ¼ jl ðxi Þ  lB ðxi Þjp þ jmA ðxi Þ  mB ðxi Þjp :
Definition 4 [17] For two arbitrary q-ROF numbers =1 ¼ 2n x2X A 2n x2X

ðl1 ; m1 Þ and =2 ¼ ðl2 ; m2 Þ and q  1. Then, ð6Þ


(1) If SF ð=1 Þ\SF ð=2 Þ, then =1 =2 ; Furthermore, we can also attain different special
(2) If SF ð=1 Þ ¼ SF ð=2 Þ, then, cases of Minkowski distance by taking diverse val-
• If AF ð=1 Þ [ AF ð=2 Þ, then =1 =2 ; ues of p, displayed as:
• If AF ð=1 Þ ¼ AF ð=2 Þ, then =1 =2 . (a) when p ¼ 1, Minkowski distance reduces to
  Hamming distance, namely,
Definition 5 [17] Let =j ¼ lj ; mj ðj ¼ 1ð1ÞnÞ be a family
1 X
of q-ROF numbers. - ¼ ð-1 ; -2 ; . . .; -n ÞT be the weight ddh ð A; BÞ ¼ jl ðxi Þ  lB ðxi Þj
P 2n x2X A
vector of =j with -j 2 ½0; 1 and nj¼1 -j ¼ 1. Then the q- ! ð7Þ
ROFWA and q-ROFWG operators are defined as follows: 1 X
þ jmA ðxi Þ  mB ðxi Þj :
q-ROFWAð=1 ; =2 ; . . .; =n Þ 2n x2X
0vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1
u Y n  Y
n   (b) when p ¼ 2, Minkowski distance reduces to
u   
q -j - ð4Þ
¼ @ t1  mj j A;
q
1  lj ; Euclid distance, namely,
j¼1 j¼1

123
Y. Liu et al.: Prospect Theory-Based q-Rung...

1 X Table 1 Different distances A A


dde ðA; BÞ ¼ j l ð x i Þ  lB ð x i Þ j 2 between two q-ROF sets B C
2n x2X A
!1=2 ddh 0.068 0.083
1 X 2 dde 0.069 0.084
þ jmA ðxi Þ  mB ðxi Þj :
2n x2X dph 0.15 0.15
ð8Þ dle 0.099 0.0998
dcd 0.055 0.056
(c) when p ¼ 1, Minkowski distance reduces to
Chebyshev distance, namely,
ddc ðA; BÞ ¼ maxx2X fjlA ðxi Þ  lB ðxi Þj; jmA ðxi Þ  mB ðxi Þjg: X ¼ fx1 ; x2 g. According to aforesaid distances under
q-ROF environment, the results are listed in Table 2.
ð9Þ
As can be seen from the discussion above that the
(2) Peng [25] introduced another form of distance which mentioned-above distances under q-ROF environments
are different from above-mentioned Eq. (6). have the following shortcomings, that is, some extant dis-
1 X tances can not distinguish two q-ROF sets and these dis-
dph ðA; BÞ ¼ ðjðlA ðxi ÞÞq ðlB ðxi ÞÞq j þ jðmA ðxi ÞÞq tances is insensitive to uncertainty.
2n x2X
ðmB ðxi ÞÞq j þ jðpA ðxi ÞÞq ðpB ðxi ÞÞq jÞ: 3.2 Lance Distance for q-ROF Sets
ð10Þ
To overcome the mention-above shortcomings, a novel
(3) Liu [18] introduced another Euclidean distance for distance (named Lance distance) is introduced in this part.
q-ROF information, that is, The classical Lance distance was first proposed by Lance
1 X 2
and Williams [44], because it is independent of the units of
dle ðA; BÞ ¼ jðlA ðxi ÞÞq ðlB ððxi ÞÞÞq j each variable, the distance between data is measured in the
2n x2X
1=2 form of ratios, and because the ratio is less affected by
þjðmA ðxi ÞÞq ðmB ðxi ÞÞq j
2
: extreme values, it is less sensitive to bias data. For data
with large attribute bias, the distance calculated by the
ð11Þ Lance distance formula is better than other distances, so it
Based on Eq. (10) and cosine similarity between two is a common method for measuring distance in data
q-ROF numbers, the cosine distance between two q- analysis.
ROF sets are defined as follows: Definition 6 Let A ¼ fðlA ðxi Þ; mA ðxi ÞÞjxi 2 Xg, B ¼
1  CðA; BÞ þ dle fðlB ðxi Þ; mB ðxi ÞÞjxi 2 Xg are two q-ROF sets on universal
dcd ðA; BÞ ¼ ; ð12Þ
2 X ¼ fx1 ; . . .; xn g. The Lance distance between A and B is
defined as follows,
where !
1X q q
ðlA ðxi ÞÞ ðlB ðxi ÞÞ þðmA ðxi ÞÞ ðmB ðxi ÞÞ q q 1 Xn
jðlA ðxi ÞÞq ðlB ðxi ÞÞq j þ jðmA ðxi ÞÞq ðmB ðxi ÞÞq j
CðA; BÞ ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffiqffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi : dLance ¼
n x2X n i¼1 ðlA ðxi ÞÞq þðlB ðxi ÞÞq þðmA ðxi ÞÞq þðmB ðxi ÞÞq
ðlA ðxi ÞÞ2q þðlA ðxi ÞÞ2q ðmB ðxi ÞÞ2q þðlB ðxi ÞÞ2q
ð13Þ
Now, an example will be taken to show some
shortages of the above-mentioned distances under
the q-ROF environment. According to the definition of Lance distance measure,
the properties of Lance distance can be derived easily:
Example 1 Let A ¼ fð0:714; 0:64Þ; ð0:64; 0:557Þg,
B ¼ fð0:64; 0:71Þ; ð0:714; 0:641Þg, C ¼ fð0:781; 0:557Þ; Table 2 Different distances A A
ð0:557; 0:458Þg be three q-ROFSs (when q ¼ 2) on the between two q-ROF sets B C
universe X ¼ fx1 ; x2 g. According to aforesaid distances
under q-ROF environment, the results are listed in Table 1. ddh 0.1 0.098
dde 0.105 0.12
Example 2 Let A ¼ fð0:71; 0:77Þ; ð0:87; 0:37Þg, dph 0.164 0.197
B ¼ fð0:58; 0:83Þ; ð0:79; 0:5Þg, C ¼ fð0:83; 0:72Þ; ð0:89; dle 0.135 0.135
0:57Þg be three q-ROFSs (when q ¼ 3) on the universe dcd 0.085 0.085

123
International Journal of Fuzzy Systems

Theorem 1 Let A, B, C be three q-ROF sets defined in X. Table 3 Lance distance A A


Then dLance satisfies the following conditions: between two q-ROFSs (q ¼ 2) B C

(1) 0  dLance ðA; BÞ  1; dLance 0.083 0.118


(2) dLance ðA; BÞ ¼ 0 if and only if A ¼ B;
(3) dLance ðA; BÞ ¼ dLance ðB; AÞ.
In Example 1, Lance distance is employed to recalculate
Table 4 Lance distance A A
the distance between A and B, A and C, result listed in between two q-ROFSs (q ¼ 3) B C
Tables 3 and 4:
As discussed above, the proposed distances have at least dLance 0.178 0.141
two aspect merits as follows:
(1) the proposed Lance distance can distinguish two
q-ROF sets, while others fail to do so.
(2) the proposed Lance distance is highly sensitive to I1 I2 ... quad I n
0   l   l 1
uncertainty. A1 l11 l ; m11 l l12 ; m12 l ... l ; m1n l
  B l   l   1n l 
l C
¼ A2 B l ;m l l22 ; m22 l ... l2n ; m2n C
ð lÞ
(3) the proposed Lance distance can overcome the Rl ¼ aij B 21 21 C
m n B
.. B .. .. .. C;
influence of biased data on measurement results in . @ . . . C
A
 l   l   l 
the form of ratios, making it more stable. Am lm1 ; mm1 l lm2 ; mm2 l ... lmn ; mmn l
 
where alij ¼ llij ; mlij ðl ¼ 1; 2; . . .; kÞ is a q-ROF number,
 q  q
4 The Presented q-ROF MCGDM Method llij ; mlij 2 ½0; 1 and 0  llij þ mlij  1.
In this MCGDM problem, the weights of experts and
In this section, we propose a novel hybrid MCGDM indicators are unknown completely. Therefore, to address
methodology with q-ROF information based on BWM, this kind of decision-making problem, the weights of
CRITIC, and extended TODIM method. Concretely, an experts and indicators should be determined first.
extended TOPSIS method via using the proposed Lance
distance measure is presented to identify the weight of 4.2 Determination of Experts’ Weights Based
experts. Then then the improved q-ROF BWM and on Extended q-ROF TOPSIS
CRITIC methods are employed to compute the subjective
and objective weight of criteria, respectively. Lastly, the In the MCGDM problem, the importance of experts is
novel extended TODIM approach is propounded based on an important factor in the decision process due to dif-
the proposed Lance distance measure for determining the ferent experts play different roles. The weight of experts
ranking of schemes. The flowchart of the presented q-ROF- is the key problem to be solved in group decision-
TODIM methodology can be depicted in Fig. 1. making. In general, the closer an expert’s opinion is to
that of a group, the more important the information that
4.1 Generally Description of MCGDM Problem the expert provides. Therefore, based on reference [26],
the positive ideal decision (PID) matrix, left-negative
The MCGDM model is always used to find the best option ideal decision (LNID) matrix, and right-negative ideal
from a finite set of schemes concerning a group of prede- decision (RNID) matrix are also built, and an improved
fined indicators based on the preference of multiple q-ROF TOPSIS method is proposed to determine the
experts. An MCGDM problem with q-ROF information experts’ weights. The implementation processes are
consists of the following three aspects: a set of schemes are expounded as follows:
denoted as A ¼ fA1 ; A2 ; . . .; Am g, a collection of indica-  
Step 1 Construct PID RP ¼ uij P ; vij P ; where uij P ¼
tors is signified as I ¼ fI 1 ; I 2 ; . . .; I n g and group experts P P m n
k k
indicated as E ¼ fE 1 ; E 2 ; . . .; E k g. Then, the kth expert 1
k
l P 1
l¼1 uij and vij ¼ k
l P
l¼1 vij . R is the average
provides the evaluation information of schemes matrix of the individual preference decision
Ai ði ¼ 1; 2; . . .; mÞ over indicator I j ð j ¼ 1; 2; . . .; nÞ and matrices of all experts and reflected the views of
construct the q-ROF decision making matrix, displayed as the group to some extent.

123
Y. Liu et al.: Prospect Theory-Based q-Rung...

1.5
v1
v3
v5
1

0.5

0
Vi(x)(i=1,3,5)

−0.5

−1

−1.5

−2

−2.5
−1 −0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1
X

Fig. 1 Comparisons on value functions (vi ðxÞði ¼ 1; 3; 5Þ) of prospect theory


  4.3.1 Determination of the Criteria Subjective Weights
Step 2 Define LNID RL ¼ uij L ; vij L m n and RNID RR
  via q-ROF Group Individual BWM
¼ uij R ; vij R m n , where uij L ¼ min1  l  k fuij l g;
vij L ¼ max1  l  k fvij l g, uij R ¼ max1  l  k fuij l g; In this subsection, the q-ROF-GIBWM method will be
vij R ¼ min1  l  k fvij l g. employed to determine the subjective weights of the cri-
 
Step 3 Calculate Lance distances dLance Rl ; RP , dLance teria. First of all, multiplicative consistency will be intro-
 l L  
R ; R and dLance Rl ; RR . duced to define fuzzy preference relations before giving the
Step 4 Calculate the closeness index to the ideal solution group q-ROF-BWM method. The multiplicative consis-
of expert E l by the following equation: tency for q-ROF preference relations is defined as follows:
     
dLance RL ; Rl þ dLance RR ; Rl Definition 7 Let A ¼ aij n n be q-ROF preference
cl ¼
dLance ðRl ; RP Þ þ dLance ðRL ; Rl Þ þ dLance ðRR ; Rl Þ ð14Þ
relations (q-ROFPR). Then, A can become a multiplicative
ðl ¼ 1; 2; . . .; kÞ:
consistency if
aBi aij ajB ¼ aiB aji aBj ; ð16Þ
Step 5 Calculate weight wl of expert E l by the following
equation: aij ajW aWi ¼ aji aWj aiW ; ð17Þ
cl
wl ¼ Pk ðl ¼ 1; 2; . . .; kÞ: ð15Þ where B  i  j  W.
l¼1 cl
If q-ROFPR with multiplicative consistency, then aBj
and the weights kB ; kj are relative, aWj and the weights
4.3 Determination of the Attributes’ Weights kW ; kj are also relative, therefore
kB
aBj ¼ ; ð18Þ
The importance of criteria is an essential element during kB þ kj
the procedure of uncertain decision analysis to attain a
kj
more reasonable and accurate decision outcome. In this ajW ¼ ð19Þ
study, the combination weight determination method is k j þ kW
adopted to work out the synthetic weight of criteria, Above two formulas will no longer hold if the q-ROF
wherein the subjective weights are estimated with the aid preference relation is inconsistent.
of the q-ROF GIBWM approach, and the objective weights In the classical GIBWM method, Hafezalkotob [45, 46]
are estimated by means of the q-ROF CRITIC approach. suggested the senior expert assess the expertise degree of
The detailed procedures are illustrated below. panel members. Now we extend the classic GIBWM idea

123
International Journal of Fuzzy Systems

to the q-ROF environment and put forward a novel criteria Stage 2 q-ROF BWM-based individual decision-making
determination method named q-ROFGIBWM. The imple- process of the senior expert.
mentation process is illustrated as follows: The senior experts assign the best attribute and
employ a q-ROF number to express his/her
Stage 1 q-ROF BWM-based expert panel’s subjective
preference, the Best-to-others vector is written as
weights
B0 ¼ ðB0B1 ; B0B2 ; . . .; B0Bn Þ; Meanwhile, the senior
The experts’ weights ðk1 ; k2 ; . . .; kk Þ can be
experts assign the worst attribute and employ a
addressed by employing the q-ROF BWM
q-ROF number to express his/her preference, the
method, the implementation process is as
others-to-worst vector is written as
follows:
W 0 ¼ ðW1W0
; W2W 0 0
; . . .; WnW Þ. The optimal
Step 1.1 Suggest a ‘‘senior expert (E 0 )’’ and attribute weight ðx1 ; x2 ; . . .; xn Þ can be obtained
decision panel fE 1 ; E 2 ; . . .; E k g; by Model 2.
Step 1.2 According to the expertise and
knowledge, the best expert B is Model 2
assigned by the senior expert the and min f
q-ROF best-to-others vector of the 8
>
> xB
expertise degrees of the panel is >
>  SFðB0Bj Þ \f; for all j
>
> x B þ xj
articulated as >
>
< xj ð21Þ
0
Be ¼ ðaB1 ¼ ðuB1 ; vB1 Þ; s:t:  SFðWjW Þ \f; for all j
>
> x j þ xW
>
> Pn
aB2 ¼ ðuB2 ; vB2 Þ; . . .; >
>
>
> j¼1 xj ¼ 1;
aBk ¼ ðuBk ; vBk ÞÞ :
xj  0; j ¼ 1; . . .; n:
The worst expert W is assigned by
the senior expert and the q-ROF best-
Stage 3 Group q-ROF BWM-based decision-making
to-others vector of the expertise
process of the expert panel.
degrees of the panel is articulated as
The i-experts assign the best attribute and
e ¼ ða1W ¼ ðu1W ; v1W Þ;
W employ a q-ROF number to express his/her
a2W ¼ ðu2W ; v2W Þ; . . .; preference, the Best-to-others vector is written as
akW ¼ ðukW ; vkW ÞÞ: Bi ¼ ðBiB1 ; BiB2 ; . . .; BiBn Þ; Meanwhile, the i-
experts assign the worst attribute and employ a
q-ROF number to express his/her preference, the
Step 1.3 Solve the optimal weights others-to-worst vector is written as
 
k1 ; k2 ; . . .; kk for the panel by W i ¼ ðW1Wi
; W2W i i
; . . .; WnW Þði ¼ 1; 2; . . .; kÞ.
Model 1. The optimal attribute weight ðw1 ; w2 ; . . .; wn Þ
Based on the above formulas, the can be obtained by Model 3.
optimal weights can be addressed by
Model 3
the Model 1
X
k

Model 1 min ki f i
i¼1
8
min  > wB
8 >
>  SFðBiBj Þ \fi ;
kB >
>
>
> >
> w j þ wB
>
>  SFðaBj Þ \; for all j > ð22Þ
>
> k B þ kj
< wj
>
> s:t:  SFðWjW i
Þ \fi ;
< kj ð20Þ >
> wW þ wj
s:t:  SFðajW Þ \; for all j >
> P
> k þ k > n
>
> j W
Pn
>
>
> j¼1 wj ¼ 1;
>
> :
>
>
> j¼1 kj ¼ 1; wj  0; j ¼ 1; . . .; n:
:
kj  0; j ¼ 1; . . .; n:
  where j ¼ 1; 2; . . .; n and i ¼ 1; 2; . . .; k.
The optimal weights k1 ; k2 ; . . .; kk will be obtained and Combining Model 2 and Model 3, it follows that the
 represents the consistency level of q-ROFPR. following model to obtain the subjective weights

123
Y. Liu et al.: Prospect Theory-Based q-Rung...

ðxsub sub sub


1 ; x2 ; . . .; xn Þ of attributes by employing q-ROF
Step 5 Compute the correlation coefficient between the
GIBWM method: jth and kth criteria.
Model 4 P m  0  0 
sij  sj sik  sk
X
k i¼1 ð27Þ
min z ¼ pf þ ð1  pÞ ki f i qjk ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Pm  0  2 Pm  0 2 ;
i¼1 i¼1 sij  sj i¼1 sik  sk
8
> xB
>
>  SFðBiBj Þ \fi ;
>
> x j þ xB
>
> Step 6 Compute the amount of information of criteria by
>
> xj
>
> i means of Eq. (28).
>
>  SFcðWjW Þ \fi ;
>
> x W þ x j rj
>
> ð23Þ cj ¼ P
< xB n  
s:t:  SFðB0Bj Þ \f; for all j 1q ð28Þ
jk
>
> x B þ xj k¼1
>
>
>
> xj
>
>  SFðWjW 0
Þ \f; for all j
>
> x þ xW
>
> j Step 7 Find the objective weight of criteria by Eq. (29).
>
> P
> nj¼1 xj ¼ 1;
> cj
>
: xobj
j ¼ P ;
xj  0; j ¼ 1; . . .; n: n
ð29Þ
cj
j¼1
4.3.2 q-ROF CRITIC-Based Objective Weights of Criteria

In this section, the q-ROF-CRITIC method will be


employed to determine the objective weights of attributes. 4.3.3 Modified Combinative Weigh (MCW)
 
Step 1 Aggregate the Rl ¼ al ðl ¼ 1; 2; . . .; kÞ to a Based on the subjective weight determination model q-
ij
m n   ROF group individual BWM and subjective weight deter-
comprehensive decision matrix R ¼ aij m n by
mination model q-ROF CRITIC method, we employ an
Eq. (24).
  improved combinative weighting method to determine the
aij ¼ q-ROFWA a1ij ; a2ij ; . . .; akij weight vector fx1 ; x2 ; . . .; xn g of the criteria by:
0vu
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1
u Y k   q w l Y k  w ð24Þ h   e iþe
1

t xobj sub
q l
¼@ 1 1  llij ; ij m l A j x j
xj ¼ 1 ; ð30Þ
Pn h obj   sub e iþe
l¼1 l¼1

j¼1 xj xj

Step 2 Compute the score of all q-ROF number aij in where ; e 2 ½0; 1 is all positive real numbers.
 
R ¼ aij m n as well as denoted as score matrix
S ¼ ðsij Þm n . 4.4 Extended TODIM Approach Based on Prospect
0 0
Step 3 Attain the normalize score matrix S ¼ ðsij Þ by the Theory
following ways:
8 In prospect theory, the prospect value consists of the value
>
> sij  min sij
>
> 1im
; if I j is benefit type; function and decision weight function. The value function
>
>
< max sij  min sij
0 1im 1im given by Tversky and Kahneman [47] well reflects the
sij ¼
max sij  sij
ð25Þ
>
> characteristic that decision makers are risk averse when
>
> 1im
>
> ; if I j is cost type:
: max sij  min sij facing gains and risk preference when facing losses. The
1im 1im
specific expression is as follows:
xa x0
Step 4 Compute the indicator’s standard deviations by v1 ðxÞ ¼ ð31Þ
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi kðxÞb x\0
uP
um  0 2
u sij  sj Where x represents the difference between the decision
t ð26Þ
rj ¼ i¼1 ; scheme and the reference point, and x  0 represents the
m
Pm 0 income; When x\0 means loss. a and b ( 0\a; b\1) are
sij
where sj ¼ i¼1 . risk preference and risk aversion coefficients, respectively.
n
The larger a; b are, the more adventurous the decision-
maker is. k is the loss avoidance coefficient, and k [ 1

123
International Journal of Fuzzy Systems

means that the decision-maker is more sensitive to losses. It follows from Figs. 2 and 3 that the changes of the new
It should be noted that the value function is not bounded. value function v5 ðxÞ is more evident than other value
Therefore, the bounded value functions for modeling pro- functions. More importantly, the loss aversion expressed by
spect theory can be found, the exponential model is one of V5 ðxÞ is stronger than that of other value functions. At the
the bounded mathematical models for prospect theory. The same time, the trend of the prospect value with the change
exponential function can better capture the whole risk- of the parameter n can be obtained from Fig. 4.
aversion behaviors, Rieger and Bui [48] suggested the Now, according to v5 ðxÞ and Ren [33], the q-ROF
following exponential value function: TODIM approach is improved and listed as follows:
1  eax x0 Step 1 Normalized comprehensive decision making
v2 ðxÞ ¼   ð32Þ  
k 1  ebx x\0 matrix R and obtained R~ ¼ r~ij m n ;
Step 2 Calculate the function /j ðAi ; At Þ of Ai over At
In order to provide a good prospect value judgment ability,
Hao et al. [49] suggested a natural logarithm-based value under criteria I j by the following equation:
8   
function: >
> xj ln 1 þ 10qdLance r~ij ; r~tj ;
<
/j ðAi ; At Þ ¼ 0; ð36Þ
lnð1 þ axÞ x0 >
>   
v3 ðxÞ ¼ ð33Þ :
 xj kln 1 þ 10qdLance r~tj ; r~ij ;
klnð1  bxÞ x\0
 
where dLance r~ij ; r~tj is the Lance distance between
In value functions vi ðxÞði ¼ 1; 2; 3Þ, the k ¼ 2:25 and a ¼
r~ij and r~tj .
b ¼ 0:88 are empirically suggested by Tverky and Kah-
neman [47]. Step 3 Calculated dominance ratio matrix /ðAi Þ of
Based on exponential and logarithm value function, two scheme Ai by the following equation:
 
new exponential and logarithm value function [50] is /ðAi Þ ¼ /j ðAi ; At Þ m n : ð37Þ
defined as follows:
(
1  10nx x0 Step 4 Determine the total dominance ration UðAi Þ of
v4 ðxÞ ¼ ð34Þ
nx
kð1  10 Þ x\0 alternative Ai by using the following formulation:
Xm X n
and UðAi Þ ¼ / j ð Ai ; At Þ ð38Þ
t¼1 j¼1
lnð1 þ 10nxÞ x0
v5 ðxÞ ¼ ð35Þ
klnð1  10nxÞ x\0
Step 5 Ranking the schemes according to UðAi Þ.
To depict the difference between value functions and the
intuitive comparison can be found in Figs. 2 and 3.

Fig. 2 Comparisons on value functions (vi ðxÞði ¼ 2; 4Þ)of prospect theory

123
Y. Liu et al.: Prospect Theory-Based q-Rung...

2.25
ξ=1
ξ=2
1.75
ξ=3
ξ=4
1.25 ξ=5

0.75

0.25
V5(x)

−0.25

−0.75

−1.25

−1.75

−2.25
−1 −0.8 −0.6 −0.4 −0.2 0 0.2 0.4 0.6 0.8 1
X

Fig. 3 Comparisons on value functions v5 ðxÞ with different parameters

From the above decision algorithm, we can obtain the


implementation flowchart of the proposed decision
methodology as Fig. 4.

5 Case Study

In this section, an illustrative example that assesses the


investment risk of renewable energy project are adopted to
illustrate the applicability of the propounded methodology.
Subsequently, the comparison and sensitivity analysis are
also implemented to expound the strengths and robustness
of our methodology, respectively.

5.1 Background Introduction

With the increasing depletion of fossil fuels, vigorously


developing renewable energy has become a key focus of
environmental departments around the world. To achieve
the long-term goals of carbon neutrality and carbon peak-
ing with high quality, the energy planning department and
energy technology companies will be committed to
investing in and developing different renewable energy
projects to alleviate the challenges posed by energy
shortage. A certain energy technology company plans to
invest in offshore wind power projects this year. The
company’s planning department selects four potential
investment projects denoted as fA1 ; A2 ; A3 ; A4 g through Fig. 4 Implementation flowchart of the propounded decision
framework
public bidding to evaluate their investment risks. The
company first invited five experts denoted as scholars from
invited experts are denoted as E 1 ; E 2 ; E 3 ; E 4 ; E 5 . After dis-
the fields of energy, investment, and project management
cussion, the evaluation committee selected four criteria:
to form a decision-making evaluation committee, the
technology (I 1 ), economy (I 2 ), environment (I 3 ), and

123
International Journal of Fuzzy Systems

Table 5 Explanation of the assessment criteria I i ði ¼ 1; 2; 3; 4Þ Table 6 Individual q-ROF decision matrices given by
E k ðk ¼ 1; 2; 3; 4; 5Þ
Criteria Explanation
Experts I1 I2 I3 I4
Technical criteria It refers to the influencing factors related to
(I 1 ) science and technology involved in the A1 (0.6, 0.2) (0.5, 0.1) (0.5, 0.3) (0.6, 0.3)
evaluation process of renewable energy A2 (0.7, 0.2) (0.6, 0.2) (0.4, 0.1) (0.4, 0.2)
projects, including technological maturity,
E1 A3 (0.5, 0.3) (0.5, 0.3) (0.7, 0.1) (0.4, 0.3)
technological research and development
capabilities, technological performance, and A4 (0.8, 0.1) (0.7, 0.2) (0.4, 0.3) (0.6, 0.1)
the ability to transform achievements into A1 (0.6, 0.1) (0.4, 0.5) (0.5, 0.2) (0.5, 0.3)
technologies A2 (0.7, 0.2) (0.5, 0.1) (0.7, 0.2) (0.4, 0.3)
Economic criteria It refers to the economic and financial factors E2 A3 (0.5, 0.3) (0.6, 0.3) (0.6, 0.3) (0.5, 0.4)
(I 2 ) involved in the evaluation process of
renewable energy projects, mainly involving A4 (0.5, 0.2) (0.4, 0.3) (0.5, 0.1) (0.5, 0.3)
the profitability, debt-paying ability, A1 (0.5, 0.3) (0.4, 0.4) (0.7, 0.2) (0.4, 0.2)
operational ability, and development debt- A2 (0.7, 0.1) (0.7, 0.1) (0.6, 0.3) (0.4, 0.1)
paying ability of the project enterprise
E3 A3 (0.5, 0.2) (0.4, 0.3) (0.5, 0.2) (0.6, 0.3)
Environment It refers to the factors related to environmental
A4 (0.6, 0.3) (0.5, 0.3) (0.5, 0.4) (0.5, 0.2)
criteria (I 3 ) sustainability involved in the evaluation
process of renewable energy projects, mainly A1 (0.5, 0.4) (0.5, 0.3) (0.6, 0.2) (0.5, 0.4)
including environmental carrying capacity, A2 (0.7, 0.3) (0.7, 0.2) (0.6, 0.2) (0.6, 0.2)
environmental resources, environmental E4 A3 (0.5, 0.4) (0.6, 0.4) (0.6, 0.2) (0.5, 0.3)
coordination, and environmental management
capacity A4 (0.8, 0.2) (0.7, 0.2) (0.4, 0.2) (0.5, 0.2)
Social criteria It refers to the factors related to social A1 (0.6, 0.3) (0.6, 0.4) (0.8, 0.2) (0.4, 0.2)
(I 4 ) development involved in the evaluation A2 (0.8, 0.1) (0.7, 0.2) (0.6, 0.3) (0.4, 0.1)
process of renewable energy projects, mainly E5 A3 (0.5, 0.2) (0.4, 0.3) (0.5, 0.2) (0.6, 0.3)
including social carrying capacity, social
A4 (0.7, 0.3) (0.5, 0.4) (0.5, 0.4) (0.5, 0.2)
profits, and public support

Step 3-1 Determine the subject weight of criteria by the


q-ROF-BWM method. The detailed process are
society (I 4 ) to comprehensively evaluate the investment
listed as follows:
risks of the above four projects, the detailed explain of the
Determine expert panel weights by the q-ROF-
mentioned four criteria are expounded in Table 5.
BWM method. Firstly, according to the expert
Based on the considered criteria, experts provide their
knowledge and expert expertise, specifically a
evaluations of the investment projects with the aid of
senior expert E0 , and determine the best expert
q-ROF number to model the uncertainty of their cognition,
E 3 and worst expert E 2 by a senior expert.
the corresponding evaluation information of experts is
Secondly, we employ the fuzzy preference scale
collected in Table 6.
to express the best-to others and others to worst
vectors of experts’ expertise degrees and listed
5.2 Decision-Making Process for Selecting the Best
Table 7.
Project
Thirdly, the BWM method will be employed to
determine the experts’ subjective weights, the
Step 1 Since all attributes are benefit-type, it is not nec-
model is listed as follows:
essarily normalized.
min f
Step 2 Determining the 8
>
> w3 w3 w3
>
>  0:815 \f ;  0:905 \f ;  0:635 \f
> w1 þ w3 w2 þ w3 w4 þ w3
experts weights. According to the Eqs. (14), (15) >
>
>
>
>
< w3
 0:725 \f ;
w1
 0:59 \f ;
w4
 0:815 \f ð39Þ
and take q ¼ 2, we obtain the expert weight vector s:t:
>
>
>
w5 þ w3 w1 þ w2 w4 þ w2
>
> w 5
>  0:635 \f
(0.195, 0.191, 0.209, 0.205, 0.2). >
>
>
>
:
w4 þ w5
w1 þ w2 þ w3 þ w4 þ w5 ¼ 1:
Step 3 Determining the weight of criteria. The proposed
combinative weight method will be adopted to Solving the program problem by LINGO 11,
address the criteria weight. the results are listed as follows:
w1 ¼ 0:107; w2 ¼ 0:0636; w3 ¼ 0:4178;
w4 ¼ 0:2811; w5 ¼ 0:1304;
and the consistency level f ¼ 0:037.

123
Y. Liu et al.: Prospect Theory-Based q-Rung...

Table 7 Expertise degree of Senior expert Best and worst expert q-ROF preference
decision experts
E1 E2 E3 E4 E5

E0 Best: E 3 (0.8, 0.1) (0.9, 0) (0.5, 0.5) (0.6, 0.3) (0.7, 0.2)
Worst: E 2 (0.55, 0.35) (0.5, 0.5) (0.9, 0.5) (0.8, 0.1) (0.6, 0.3)

Step 3-2 Determine the attribute’s subject weight by the Step 3-3 Determine the attribute’s objective weight by
group individual q-ROF BWM method. The the q-ROF BWM method. The detailed process
detailed process is listed as follows: is listed as follows:
Firstly, all experts determine the best attributes Firstly: Aggregate the Rl ðl ¼ 1; 2; . . .; 5Þ to a
and worst attributes. Secondly, experts employ comprehensive decision matrix R ¼ ðuij ; vij Þ4 4
the fuzzy preference scale to express the best- by Eq. (24) and the outcomes are listed in
to-others and others-to-worst vectors of Table 9.
preference degrees of attributes listed in Secondly: By the Score function of q-ROF
Table 8. number, compute the score of all q-ROFN aij in
Thirdly, the BWM method will be employed to R ¼ ðuij ; vij Þ4 4 as well as denoted as score
determine the experts’ subjective weights, the matrix S ¼ ðsij Þ 4 4 and listed in Table 10.
model listed as follows: Thirdly: Attain the normalize score matrix S ¼
0

0
ðsij Þ by Eq. (25), the result is shown in
min j ¼ px0 þ ð1  pÞð0:107x1 þ 0:0636x2 þ 0:4178x3 þ 0:2811x4 þ 0:1304x5 Þ
s:t:
8
>
> 12 12 12
Table 11.
>
>  0:905 \x0 ;  0:635 \x0 ;  0:725 \x0
>
>
>
>
>
>
>
>
11 þ 12
11
12 þ 13

 0:815 \x0 ;
14
12 þ 14

 0:635 \x0
Fourth: Compute the attributes’ standard
>
>
>
> 11 þ 13 11 þ 14
>
>
>
>
>
>
>
12
 0:59 \x1 ;
12
 0:905 \x1 ;
12
 0:725 \x1
deviations by Eq. (26). And the results are listed
> 11 þ 12
> 12 þ 13 12 þ 14
>
>
>
>
>
>
>
>
>
11
11 þ 13
 0:68 \x 1 ;
14
13 þ 14
 0:635 \x1 as r1 ¼ 0:611; r2 ¼ 0:405; r3 ¼ 0:369;
>
>
>
>
>
>
>
>
>
>
12
1 þ 12
 0:59 \x2 ;
12
12 þ 13
 0:905 \x2 ;
12
12 þ 14
 0:725 \x2 r4 ¼ 0:316.
>
> 1
>
>
>
>
>
>
>
<
11
11 þ 13
 0:725 \x2 ;
14
13 þ 14
 0:68 \x2
ð40Þ Fifth: Compute the correlation coefficient
12 12 12
>
>
>
> 11 þ 12
>
>
>
 0:635 \x3 ;
12 þ 13
 0:905 \x3 ;
12 þ 14
 0:725 \x3 between indicator pair jth and kth by means of
>
> 11 14
>
>
>
>
>
>
>
>
11 þ 13
 0:725 \x3 ;
13 þ 14
 0:68 \x3
Eq. (27) and the results are listed in Table 12.
>
> 12 12 12
>
>  0:635 \x4 ;  0:815 \x4 ;  0:68 \x4
>
>
>
> 1
>
>
>
1 þ 12
11
12 þ 13
14
12 þ 14 Sixth: The objective weight can be computed
>
>  0:725 \x4 ;  0:68 \x4
>
>
>
>
>
>
>
>
> 12
11 þ 13
12
13 þ 14
12
by Eqs. (28), (29) and the results are listed as
>
>  0:59 \x5 ;  0:815 \x5 ;  0:68 \x5
>
xobj obj obj
> 11 þ 12 12 þ 13 12 þ 14

1 ¼ 0:258; x2 ¼ 0:172; x3 ¼ 0:343;


>
>
>
> 11 14
>
>  0:725 \x5 ;  0:675 \x5
>
> 1 þ 1 1 þ 1
>
> 1 3 3 4
:
xobj
11 þ 12 þ 13 þ 14 ¼ 1:
4 ¼ 0:227:
Step 3-4 Determine modified combinative weights
Solving the program problem by LINGO 11, ð-1 ; -2 ; . . .; -4 Þ of the criteria by Eq. (30).
the results are listed as follows: Taking  ¼ e ¼ 1, and obtained the MCW as
x1 ¼ 0:28; x2 ¼ 0:328; x3 ¼ 0:182;
11 ¼ 0:249; 12 ¼ 0:512;
x4 ¼ 0:211:
13 ¼ 0:078; 14 ¼ 0:160 Step 4 Rank the alternatives by extended TODIM
and the consistency level j ¼ 0:069. We approach based on prospect theory.
denote the subjective weight as follows:
xsub sub Step 4- Normalized comprehensive decision-making
1 ¼ 11 ¼ 0:249; x2 ¼ 12 ¼ 0:512;  
1 matrix R and obtained R~ ¼ r~ij 4 4 , in this
xsub sub
3 ¼ 13 ¼ 0:078; x4 ¼ 14 ¼ 0:160:
example, all attributes are benefited, so it is
necessary to normalize the comprehensive
decision-making matrix.

123
International Journal of Fuzzy Systems

Table 8 Expertise degree of Experts Best and worst criteria q-ROF fuzzy preference
decision experts
I1 I2 I3 I4

E0 Best: I 2 (0.90, 0.00) (0.50, 0.50) (0.60, 0.30) (0.70, 0.20)


Worst: I 1 (0.50, 0.50) (0.90, 0.00) (0.80, 0.10) (0.6, 0.3)
E1 Best: I 2 (0.55, 0.35) (0.5, 0.5) (0.9, 0) (0.7, 0.2)
Worst: I 3 (0.65, 0.25) (0.80, 0.10) (0.50, 0.50) (0.60, 0.30)
E2 Best: I 2 (0.55, 0.35) (0.50, 0.50) (0.90, 0.00) (0.70, 0.20)
Worst: I 3 (0.70, 0.20) (0.90, 0.00) (0.50, 0.50) (0.65, 0.25)
E3 Best: I 2 (0.6, 0.3) (0.5, 0.5) (0.9, 0) (0.7, 0.2)
Worst: I 3 (0.70, 0.20) (0.90, 0.00) (0.50, 0.50) (0.625, 0.25)
E4 Best: I 2 (0.60, 0.30) (0.50, 0.50) (0.90, 0.00) (0.70, 0.20)
Worst: I 3 (0.70, 0.20) (0.90, 0.00) (0.5, 0.50) (0.65, 0.25)
E5 Best: I 2 (0.55, 0.35) (0.50, 0.50) (0.80, 0.10) (0.65, 0.25)
Worst: I 3 (0.70, 0.20) (0.80, 0.10) (0.50, 0.50) (0.60, 0.10)

Table 9 The comprehensive I1 I2 I3 I4


decision matrix
A1 (0.563, 0.2383) (0.507, 0.300) (0.649, 0.216) (0.488, 0.270)
A2 (0.724, 0.164) (0.653, 0.152) (0.595, 0.206) (0.453, 0.163)
A3 (0.5, 0.270) (0.512, 0.318) (0.589, 0.189) (0.530, 0.317)
A4 (0.707, 0.206) (0.586, 0.2706) (0.464, 0.252) (0.522, 0.189)

Table 10 The score matrix


Step 4- Calculated dominance ratio matrix /ðAi Þ of
sj1 sj2 sj3 sj4 3 alternative Ai by Eq. (37), the results are listed as
A1 0.630 0.584 0.687 0.583 follows:
0 1
A2 0.749 0.702 0.656 0.589 0 0 0 0
B 0:783  0:465 C
A3 0.589 0.580 0.655 0.590 B  0:899 0:107 C
/ðA1 Þ ¼ B C
A4 0.728 0.635 0.576 0.617 @ 0:214 0:064 0:123  0:364 A
0:706  0:612 0:243  0:397
0 1
0:349 0:440  0:239 0:207
Step 4- Calculate the U function /j ðAi ; At Þ of Ai over B 0 C
B 0 0 0 C
2 At under attribute I j by Eq. (36). Taking q ¼ 10, /ðA2 Þ ¼ B C
@ 0:422 0:442 0:029  0:639 A
the results are listed as follows: 0:087 0:298 0:213  0:456
0 1 0 1
0 0:349  0:482 0:315 0:482  0:143  0:276 0:162
B 0:783
B 0  0:948  0:195 C
C B 0:948  0:993  0:065 0:284 C
/1 ðAi ; At Þ ¼ B B C
@ 0:214 0:422 0 0:395 A
C / ð A3 Þ ¼ B C
@ 0 0 0 0 A
0:706 0:087  0:889 0
0 1 0:889  0:629 0:216  0:375
0 0:440  0:143 0:273 0 1
B 0:899 0:315 0:273  0:546 0:177
B 0  0:993  0:669 C
C
/2 ðAi ; At Þ ¼ B C B 0:195  0:669  0:480 0:203 C
@ 0:064 0:280 A B C
0:442 0 / ð A4 Þ ¼ B C
0:612 0:298  0:629 0 @ 0:395 0:280  0:484 0:167 A
0 1
0  0:239  0:276  0:546 0 0 0 0
B 0:107 0  0:065  0:480 C
B C
/3 ðAi ; At Þ ¼ B C
@ 0:123 0:029 0  0:484 A
0:243 0:213 0:216 0
0 1
0 0:207 0:162 0:177
B 0:465 0 0:284 0:203 C
B C
/4 ðAi ; At Þ ¼ B C
@ 0:364  0:639 0 0:167 A
0:397  0:456  0:375 0

123
Y. Liu et al.: Prospect Theory-Based q-Rung...

Table 11 Normalization of score matrix parameters q. This point will be shown above
0 0 0 0 Table 13 and Fig. 5 that the ranking of alternatives
sj1 sj2 sj3 sj4
will remain unchanged.
A1 0.417 0.022 1 0 (2) Effect of parameter q changes in /-function on the
A2 1.621 0.996 0.715 0.187 results under q ¼ 2.
A3 0 0 0.714 0.214 Although the values of cross-entropy will increases
A4 1 0.447 0 0.838 as qðq [ 0Þ increases, the experts’ weights deter-
mined by cross-entropy are almost unchanged,
which leads to the same rank under different
Table 12 Correlation coefficient Matrix parameter q. This point will be shown above
Table 14 and Fig. 6 that the ranking of alternatives
qj1 qj2 qj3 qj4
will remain unchanged.
1 0.968 - 0.2978 0.260 (3) In Step 4-2, the function /j ðAi ; At Þ of Ai over At
0.968 1 - 0.259 0.209 under attribute aj is given in exponent function by
- 0.297 - 0.259 1 - 0.997 the following equation:
0.259 0.209 - 0.997 1 8  
> w 1  10 ðqdLance ðr~ij ;r~tj ÞÞ ;
>
> j
>
<
/j ðAi ; At Þ ¼ 0;
>
>  
>
>
:  w k 1  10ðqdLance ðr~ij ;r~tj ÞÞ ;
Step 4- Determine the total dominance ration UðAi Þ of j

4 alternative Ai by Eq. (38), the results are listed as ð41Þ


follows: 
where dLance r~ij ; r~tj Þ is the Lance distance between r~ij
UðA1 Þ ¼ 3:572; UðA2 Þ
and r~tj . We can obtain the final order when fixing
¼ 1:151; UðA3 Þ q ¼ 2, the parameters q and k take different values in
¼ 4:145; UðA4 Þ ¼ 0:571: the method and the final decision results are listed in
Table 15, Fig. 7 and Table 16, and Fig. 8,
respectively.
Step 4- Ranking the alternatives according to UðAi Þ, and
5 the ranks of the alternatives is A2 A4 A1
A3 .
5.4 Comparisons and Analyses

In this subsection, some comparison analyses will be car-


5.3 Effect of the Parameter Change on Results ried out with other previous methodologies.
(Method 1) The aggregation operator is an important
The results are discussed when the parameters change in information fusion tool. Therefore, we compare our pro-
this section. posed method with aggregation operator-based decision-
(1) Effect of parameter changes in cross-entropy on the making under q-ROF environment. The aggregation result
results under q ¼ 10. e 1 ¼ ð0:551; 0:259Þ; A
can be attained, displayed as: A e2 ¼
Although the values of cross-entropy will increases ð0:636; 0:166Þ; A e 4 ¼ ð0:598; 0:229Þ:
e 3 ¼ ð0:528; 0:276Þ; A
as qðq [ 0Þ increases, the experts’ weights deter- The scores of alternatives are SF ðA1 Þ ¼ 0:6181;
mined by cross-entropy are almost unchanged, SF ðA2 Þ ¼ 0:6884; SF ðA3 Þ ¼ 0:6016; SF ðA4 Þ ¼ 0:6527, so
which leads to the same rank under different

Table 13 Decision outcome for Parameter Ordering index of Ai ði ¼ 1; 2; 3Þ Ranking order


the diverse value of q
q¼1 UðA1 Þ ¼ 2:492; UðA2 Þ ¼ 1:043; UðA3 Þ ¼ 3:006; UðA4 Þ ¼ 0:59 A2 A4 A1 A3
q¼2 UðA1 Þ ¼ 3:521; UðA2 Þ ¼ 1:273; UðA3 Þ ¼ 4:325; UðA4 Þ ¼ 0:659 A2 A4 A1 A3
q¼3 UðA1 Þ ¼ 3:113; UðA2 Þ ¼ 0:811; UðA3 Þ ¼ 4:142; UðA4 Þ ¼ 0:475 A2 A4 A1 A3
q¼5 UðA1 Þ ¼ 4:854; UðA2 Þ ¼ 1:533; UðA3 Þ ¼ 6:521; UðA4 Þ ¼ 0:604 A2 A4 A1 A3
q ¼ 10 UðA1 Þ ¼ 5:334; UðA2 Þ ¼ 1:881; UðA3 Þ ¼ 8:537; UðA4 Þ ¼ 0:789 A2 A4 A1 A3

123
International Journal of Fuzzy Systems

Fig. 5 Decision outcome for the diverse value of q

Table 14 Decision outcome for Parameter Ordering index of Ai ði ¼ 1; 2; 3; 4Þ Ranking order


diverse values of q
q¼1 UðA1 Þ ¼ 0:595; UðA2 Þ ¼ 0:249; UðA3 Þ ¼ 0:755; UðA4 Þ ¼ 0:121 A2 A4 A1 A3
q¼2 UðA1 Þ ¼ 1:092; UðA2 Þ ¼ 0:434; UðA3 Þ ¼ 1:342; UðA4 Þ ¼ 0:203 A2 A4 A1 A3
q¼3 UðA1 Þ ¼ 1:519; UðA2 Þ ¼ 0:594; UðA3 Þ ¼ 1:855; UðA4 Þ ¼ 0:279 A2 A4 A1 A3
q¼4 UðA1 Þ ¼ 1:895; UðA2 Þ ¼ 0:726; UðA3 Þ ¼ 2:305; UðA4 Þ ¼ 0:346 A2 A4 A1 A3
q¼5 UðA1 Þ ¼ 2:231; UðA2 Þ ¼ 0:839; UðA3 Þ ¼ 2:705; UðA4 Þ ¼ 0:404 A2 A4 A1 A3
q ¼ 10 UðA1 Þ ¼ 3:521; UðA2 Þ ¼ 1:259; UðA3 Þ ¼ 4:257; UðA4 Þ ¼ 0:625 A2 A4 A1 A3

the order of alternatives is A2 A4 A1 A3 , which is the membership degree change in the same scheme for dif-
same with order by our proposed method. ferent experts and different attributes. For example, when
From the above comparison, the same order is obtained the non-membership degree of evaluation information of
by aggregation-based decision-making and our proposed alternative A3 concerning a1 of the expert, that is, change
method. Although the aggregation-based approach has ð1Þ
a31 ¼ ð0:5; 0:3Þ to (0.5, 0), no matter how other non-
some merits, this approach will be invalid or the reasonable membership change, the result of alternative A3 obtained
results will be obtained by the aggregation-based approach e 3 ¼ ð0:528; 0Þ; it is
by aggregation-based approach is A
under some special cases. On the one hand, if change the
obviously unreasonable in this situation. However, in our
evaluation information of the first attribute by the first
ð1Þ
proposed method, the results may be different when the
expert, that is, change a11 ¼ ð0:6; 0:2Þ to (1, 0), the non-membership degree concerning a certain
desirable alternative is A1 . If any expert changes some scheme changes.
attribute’s evaluation value of alternative Ai to (1, 0), the (Method 2) TOPSIS method. The close indexes are
desirable alternative must be Ai , this is obviously unrea- listed as follows: C ðA1 Þ ¼ 0:0706; CðA2 Þ ¼
sonable. However, in this situation, our proposed method 0:1578; CðA3 Þ ¼ 0:0446; C ðA4 Þ ¼ 0:1005. Therefore, the
can avoid the issue, for instance, if a111 ¼ ð1; 0Þ, the result ranking result is A2 A4 A1 A3 .
obtained by our proposed method will be obtained as fol- (Method 3) q-ROF EDAS method. Darko and Liang
lows: U ðA1 Þ ¼ 1:298; UðA2 Þ ¼ 0:047; UðA3 Þ ¼ [29] proposed a modified EDAS group decision method
3:998; UðA4 Þ ¼ 1:435, so the order of alternatives is under q-ROF environment. We also applied the EDAS
A2 A4 A1 A3 . On the other hand, if a non-member- approach to address our problems. The appraisal score is
ship degree of evaluation information is 0, the aggregation AS1 ¼ 0:5115; AS2 ¼ 0:5708; AS3 ¼ 0:4786;
results will not change when no matter how other non- AS4 ¼ 0:5375, so the ranking result is A2 A4 A1 A3 .

123
Y. Liu et al.: Prospect Theory-Based q-Rung...

Fig. 6 Decision outcome for the diverse value of q

Table 15 Decision outcome for Parameter Ordering index of Ai ði ¼ 1; 2; 3; 4Þ Ranking order


diverse values of q under
exponent function in Step 4-2 q¼1 UðA1 Þ ¼ 1:165; UðA2 Þ ¼ 0:463; UðA3 Þ ¼ 1:492; UðA4 Þ ¼ 0:238 A2 A4 A1 A3
q¼2 UðA1 Þ ¼ 1:884; UðA2 Þ ¼ 0:678; UðA3 Þ ¼ 2:338; UðA4 Þ ¼ 0:361 A2 A4 A1 A3
q¼3 UðA1 Þ ¼ 2:336; UðA2 Þ ¼ 0:776; UðA3 Þ ¼ 2:851; UðA4 Þ ¼ 0:424 A2 A4 A1 A3
q¼4 UðA1 Þ ¼ 2:624; UðA2 Þ ¼ 0:82; UðA3 Þ ¼ 3:184; UðA4 Þ ¼ 0:456 A2 A4 A1 A3
q¼5 UðA1 Þ ¼ 2:811; UðA2 Þ ¼ 0:839; UðA3 Þ ¼ 3:416; UðA4 Þ ¼ 474 A2 A4 A1 A3
q ¼ 10 UðA1 Þ ¼ 3:13; UðA2 Þ ¼ 0:862; UðA3 Þ ¼ 3:957; UðA4 Þ ¼ 0:527 A2 A4 A1 A3

(Method 4) q-ROF-COPRAS proposed. Krishankumar


et al. [28] proposed the COPRAS group decision method
under q-ROF environment. We also applied the COPRAS
approach to address our problems. The implementation
process is as follows:
Based on the attributes’ weights and experts’ weights
proposed method in the current work, the implementation
procedures are as follows:
(a) Aggregation individual decision making matrices
 
R~ðlÞ ¼ a~lij to a comprehensive decision-matrix
m n  
into a collective decision matrix R~ ¼ a~ij m n .
(b) Calculate the sum of the degrees of attribute for
different types. In this step, each option needs to
calculate scores for maximizing benefit indicators
and minimizing cost indicators: Fig. 7 Decision outcome for diverse values of q under exponent
function in Step 4-2
l
ai ¼  wj e
a ij ; i ¼ 1ð1Þm:
j¼1 (c) Calculate the ‘‘relative degree (RD)’’ of each alter-
n
bi ¼  wj e
a ij ; i ¼ 1ð1Þm: native (or scheme). The RDðci Þ of the ith alternative
j¼lþ1 is obtained using

123
International Journal of Fuzzy Systems

Table 16 Decision outcome for Parameter Ordering index of Ai ði ¼ 1; 2; 3; 4Þ Ranking order


diverse values of k
k ¼ 0:05 UðA1 Þ ¼ 0:4533; UðA2 Þ ¼ 0:3068; UðA3 Þ ¼ 0:3613; UðA4 Þ ¼ 0:4018 A2 A4 A1 A3
k ¼ 0:5 UðA1 Þ ¼ 0:352; UðA2 Þ ¼ 0:2942; UðA3 Þ ¼ 0:3546; UðA4 Þ ¼ 0:3933 A2 A4 A1 A3
k¼1 UðA1 Þ ¼ 0:4147; UðA2 Þ ¼ 0:2807; UðA3 Þ ¼ 0:3465; UðA4 Þ ¼ 0:3829 A2 A4 A1 A3
k¼5 UðA1 Þ ¼ 0:2982; UðA2 Þ ¼ 0:2110; UðA3 Þ ¼ 0:2806; UðA4 Þ ¼ 0:2974 A2 A4 A1 A3
k ¼ 10 UðA1 Þ ¼ 0:2334; UðA2 Þ ¼ 0:1754; UðA3 Þ ¼ 0:2268; UðA4 Þ ¼ 0:2311 A2 A4 A1 A3
k ¼ 50 UðA1 Þ ¼ 0:1540; UðA2 Þ ¼ 0:1440; UðA3 Þ ¼ 0:1506; UðA4 Þ ¼ 0:1527 A2 A4 A1 A3
k ¼ 100 UðA1 Þ ¼ 0:1516; UðA2 Þ ¼ 0:1406; UðA3 Þ ¼ 0:1432; UðA4 Þ ¼ 0:1446 A2 A4 A1 A3

Table 18 The ranking results using different approaches


Approaches Ranking order

Aggregation-based method [17] A2 A4 A1 A3


TOPSIS [51] A2 A4 A1 A3
q-ROF EDAS [29] A2 A4 A1 A3
q-ROF-COPRAS [28] A2 A4 A1 A3
q-ROF-WASPAS [30] A2 A4 A1 A3
Our proposed method A2 A4 A1 A3

Therefore, according to the UD, the ranking results


is A2 A4 A1 A3 .
Fig. 8 Decision outcome for the diverse value of k
(Method 5) q-ROF-WASPAS method. Rani and Mishra
Pm
min Sðbi Þ i¼1 Sðbi Þ [30] proposed the WASPAS group decision method under
i
c i ¼ S ð ai Þ þ P mini Sðbi Þ : ð42Þ q-ROF environment. We also applied the WASPAS
Sðbi Þ m i¼1 Sðb Þ approach to address our problems. The implementation
i

where Sðai Þ and Sðbi Þ denote the score values of ai process is as follows:
and bi , respectively. Based on the attributes’ weights and experts’ weights
(d) Calculate the ‘‘utility degree (UD)’’ of each alterna- proposed method in the current work, the rest of the
tive. By evaluating the examined alternatives with implementation is as follows:
the optimal one, the UD of each alternative is (a) Aggregation individual decision-making matrices
 
calculated using.
R~k ¼ a~kij to a comprehensive decision-matrix
c m n
di ¼ i ; i ¼ 1; 2; . . .; m: ð43Þ into a collective individual decision matrix R~ ¼
cmax  
a~ij m n .
Based upon the above-mentioned q-ROF-COPRAS (b) Estimate the weighted sum measure and the
algorithm, the UD of alternatives is obtained as weighted product measure using the following q-
follows: ROF aggregation operators:
d1 ¼ 0:898; d2 ¼ 1; d3 ¼ 0:874; d4 ¼ 0:948; ð44Þ

Table 17 The q-ROF- Alternatives ai bi Sðai Þ Sðbi Þ Qi ða ¼ 0:5Þ Ranking order


WASPAS method for ranking
alternatives A1 (0.551, 0.259) (0.542, 0.264) 0.618 0.612 0.615 3
A2 (0.636, 0.166) (0.612, 0.168) 0.688 0.673 0.681 1
A3 (0.528, 0.276) (0.525, 0.286) 0.616 0.597 0.599 4
A4 (0.598, 0.229) (0.577, 0.235) 0.653 0.639 0.646 2

123
Y. Liu et al.: Prospect Theory-Based q-Rung...

Table 19 Parameters to compare with different approaches


Indexes Approaches
Aggregation-based TOPSIS [51] q-ROF EDAS q-ROF COPRAS q-ROF WASPAS Proposed method
method [17] [29] [28] [30]

Experts’ Computed (known) Computed Computed Known Known Computed (GI-


weights (known) BWM-CRITIC)
Attributes’ Computed (known) Computed Computed q-ROF COPRAS q-ROF WASPAS Computed (extended
weights (known) TOPSIS)
Aggregation q-ROFWAO q-ROF WAO q-ROF WAO q-ROF WAO q-ROFWAO, q-ROFWAO,
process q-ROFGWO TODIM
Decision Group Group Group Group Group Group
process
Sorting A2 A4 A1 A3 A2 A4 A1 A3 A2 A4 A1 A3 A2 A4 A1 A3 A2 A4 A1 A3 A2 A4 A1 A3
Desirable A2 A2 A2 A2 A2 A2
alternative

n • There are some different q-ROF-TOPSIS methods


ai ¼  wj e
a ij ; i ¼ 1; 2; . . .; m:
j¼1 designed by taking various discrimination measures,
n all developed TOPSIS are based on different distance
bi ¼  wj e
a ij ; i ¼ 1; 2; . . .; m:
j¼l measures. Various distances have been developed in
q-ROFS, after analyzing some drawbacks of extant
(c) Find the measure of weighted aggregated sum distance measures, a novel distance measure named
product assessment using Lance measure is put forward in current work to
Qi ¼ aSðai Þ þ ð1  aÞSðbi Þ; i ¼ 1; 2; . . .; m: overcome these drawbacks and Lance distance is highly
sensitive to uncertainty.
where a 2 ½0; 1 signifies the decision coefficient, • The extant decision methods rarely consider the
Sðai Þ and Sðbi Þ denote the score values of ai and bi , incomplete rational psychology of the expert panel.
severally. Thus it is interesting to transfer the loss and gain of
(d) Determine the prioritization of the alternatives prospect theory to distance, on the basis of it, the
according to Qi . extended TODIM method considering the psychology
Based upon the above-mentioned q-ROF-WASPAS algo- behavior is developed in current work.
rithm, the computed results are listed in Table 17.
Therefore, the ranking results is A2 A4 A1 A3 .
Based on the above discussion and comparison of dif-
6 Conclusion
ferent methods, the results are summarized as follows and
listed in Table 18.
This article designed a hybrid MCGDM framework by
In light of the mentioned comparisons, the detailed
incorporating the GIBWM, CRITIC, and the extended
characteristics in different stages for the mentioned
TODIM under q-ROF circumstances for the evaluation of
approaches are compared and the outcomes are portrayed
investment risks in renewable energy projects. First of all,
in Table 19.
the q-ROF Lance distance measure was presented to make
The key merits of the proposed q-ROF-TODIM
up for the deficiencies of the existing distance measure.
methodology are discussed as follows:
Then we proposed the extended TOPSIS method by
• To handle the uncertainties that appear in MADM employing the defined q-ROF Lance distance to determine
problems, weights are very important issues in infor- the weight of experts. Afterward, the improved q-ROF-
mation fusion. In current work, experts’ weights are not GIBWM and q-ROF-CRITIC approaches were propounded
given by the experts, but calculated by extended to ascertain the subjective and objective weights of criteria,
q-ROF-TOPSIS method, a novel weights determination respectively, and the combined weight is also attained.
method under q-ROF environment; and weights of the Besides, the novel extended TODIM method based on
attributes are also not given by experts, but calculated prospect theory via the defined Lance distance was devel-
by integrating GIBWM and CRITIC method in q-ROF oped to attain the priority of investment risks. Further, a
environment. hybrid decision framework q-ROF-GIBWM-CRITIC-

123
International Journal of Fuzzy Systems

TODIM was built to evaluate the investment risk of Future research will devote to settling the mentioned
renewable energy projects. We implemented the empirical limitations. In addition, we also unfold the following
analysis and sensitivity study to confirm the practicability research: (1) The propounded decision framework will be
and stability of the propounded decision framework. The employed for other practical applications, such as sustain-
advantages and merits of the suggested decision framework ability transport [52], cold chain logistics distribution center
are verified by comparing it with the prior q-ROF decision evaluation [53], medical waste treatment scheme selection
approaches. The comparison results show that the pre- [54]. (2) The proposed Lance distance measure can be uti-
sented framework not only has strong stability but also lized to other approaches such as MABAC, Taxonomy.
considers the psychological behavior of experts during the Besides, the Lance distance measure can be extended to other
evaluation process and makes the decision results more uncertain setting such as linguistic pythagorean fuzzy set
reasonable. [55], single-valued neutrosophic set [56], picture fuzzy set
In what follows, we discuss some limitations of the [57]. (3) To further strength the reasonability of the decision
introduced approach: (1) This method fails to take into outcomes, we can take into the consensus reaching process to
account the consensus-reaching process, which means that establish other enhanced group decision models and large-
the consistency of decision experts is neglected. (2) The scale group decision approaches [58–61].
correlation relation among the fused expert information is
not been considered. (3) The assessment criteria index
system for evaluating the investment risk of renewable Appendix
energy project should be considered the aspect of sustain-
ability, low carbon and green-quality development. A list of all abbreviations in this manuscript is summarized
in Table 20.

Table 20 List of abbreviations Abbreviation Full name

TODIM An acronym in Portuguese for interactive and multi-criteria decision making


q-ROF q-Rung orthopair fuzzy
GIBWM Group individual best and worst method
CRITIC CRiteria importance through inter-criteria correlation
COPRAS COmplex PRoportional ASsessment
MABAC Multi-attributive border approximation area comparison
GRA Grey relational analysis
DEMATEL Decision making trial and evaluation laboratory
QUALIFLEX QUALItative FLEXible multiple criteria method
FS Fuzzy set
IFS Intuitionistic fuzzy set
PFS Pythagorean fuzzy set
MCDM Multi-criteria decision-making
TOPSIS Technique for order of preference by similarity to an ideal solution
VIKOR Vlsekriterijumska Optimizacija I Kompromisno Resenje
EDAS Evaluation based on distance from average solution
MCGDM Multi-criteria group decision-making
WASPAS Weighted Aggregated Sum Product ASsessment
MARCOS Measurement of Alternatives and Ranking according to COmpromise Solution
PID Positive ideal decision
LNID Left-negative ideal decision
RNID Right-negative ideal decision
q-ROFPR q-ROF preference relations
MCW Modified combinative weigh
MABAC Multi-attributive border approximation area comparison

123
Y. Liu et al.: Prospect Theory-Based q-Rung...

Acknowledgements This work was supported by Application Basic 18. Liu, P., Liu, W.: Multiple-attribute group decision-making based
Research Plan Project of Sichuan Province under Grant 2021JY0108, on power Bonferroni operators of linguistic q-rung orthopair
Scientific Research Project of Neijiang Normal University under fuzzy numbers. Int. J. Intell. Syst. 34(4), 652–689 (2019)
Grants (18TD08, 2022ZD10). 19. Liu, D., Chen, X., Peng, D.: Some cosine similarity measures and
distance measures between q-rung orthopair fuzzy sets. Int.
Data availability No data was used for the research described in the J. Intell. Syst. 34(7), 1572–1587 (2019)
article. 20. Mahmood, T., Ali, Z.: Entropy measure and TOPSIS method
based on correlation coefficient using complex q-rung orthopair
Declarations fuzzy information and its application to multi-attribute decision
making. Soft Comput. 25, 1249–1275 (2021)
Conflict of interest The authors declare that they have no conflicts of 21. Arya, V., Kumar, S.: Extended TODIM method based on VIKOR
interest. for q-rung orthopair fuzzy information measures and their
application in MAGDM problem of medical consumption prod-
ucts. Int. J. Intell. Syst. 36(11), 6837–6870 (2021)
22. Mishra, A.R., Rani, P.: A q-rung orthopair fuzzy ARAS method
References based on entropy and discrimination measures: an application of
sustainable recycling partner selection. J. Ambient Intell. Human.
1. Kul, C., Zhang, L., Solangi, Y.A.: Assessing the renewable Comput. (2021). https://doi.org/10.1007/s12652-021-03549-3
energy investment risk factors for sustainable development in 23. Mishra, A.R., Rani, P., Pamucar, D., Hezam, I.M., Saha, A.:
Turkey. J. Clean. Prod. 276, 124164 (2020) Entropy and discrimination measures based q-rung orthopair
2. Wu, Y., Wang, J., Ji, S., Song, Z.: Renewable energy investment fuzzy MULTIMOORA framework for selecting solid waste dis-
risk assessment for nations along China’s Belt & Road Initiative: posal method. Environ. Sci. Pollut. Res. 30(5), 12988–13011
an ANP-cloud model method. Energy 190, 116381 (2020) (2023)
3. Chebotareva, G., Strielkowski, W., Streimikiene, D.: Risk 24. Du, W.S.: Minkowski-type distance measures for generalized
assessment in renewable energy projects: a case of Russia. orthopair fuzzy sets. Int. J. Intell. Syst. 33(4), 802–817 (2018)
J. Clean. Prod. 269, 122110 (2020) 25. Peng, X., Liu, L.: Information measures for q-rung orthopair
4. Zhou, P., Luo, J., Cheng, F., Yüksel, S., Dinçer, H.: Analysis of fuzzy sets. Int. J. Intell. Syst. 34(8), 1795–1834 (2019)
risk priorities for renewable energy investment projects using a 26. Liao, H., Zhang, H., Zhang, C., Wu, X., Mardani, A., Al-Bar-
hybrid IT2 hesitant fuzzy decision-making approach with alpha akati, A.: A q-rung orthopair fuzzy GLDS method for investment
cuts. Energy 224, 120184 (2021) evaluation of BE angel capital in China. Technol. Econ. Dev.
5. Ilbahar, E., Kahraman, C., Cebi, S.: Risk assessment of renewable Econ. 26(1), 103–134 (2020)
energy investments: a modified failure mode and effect analysis 27. Zeng, S., Hu, Y., Xie, X.: Q-rung orthopair fuzzy weighted
based on prospect theory and intuitionistic fuzzy AHP. Energy induced logarithmic distance measures and their application in
239, 121907 (2022) multiple attribute decision making. Eng. Appl. Artif. Intell. 100,
6. Hashemizadeh, A., Ju, Y., Bamakan, S.M.H., Le, H.P.: Renew- 104167 (2021)
able energy investment risk assessment in belt and road initiative 28. Krishankumar, R., Ravichandran, K.S., Kar, S., Cavallaro, F.,
countries under uncertainty conditions. Energy 214, 118923 Zavadskas, E.K., Mardani, A.: Scientific decision framework for
(2021) evaluation of renewable energy sources under q-rung orthopair
7. Wu, Y., Hu, M., Liao, M., Liu, F., Xu, C.: Risk assessment of fuzzy set with partially known weight information. Sustainability
renewable energy-based island microgrid using the HFLTS-cloud 11(15), 4202 (2019)
model method. J. Clean. Prod. 284, 125362 (2021) 29. Darko, A.P., Liang, D.: Some q-rung orthopair fuzzy Hamacher
8. Karamoozian, A., Wu, D., Lambert, J.H., Luo, C.: Risk assess- aggregation operators and their application to multiple attribute
ment of renewable energy projects using uncertain information. group decision making with modified EDAS method. Eng. Appl.
Int. J. Energy Res. 46(13), 18079–18099 (2022) Artif. Intell. 87, 103259 (2020)
9. Li, W., Cui, Z., Tang, X.: Risk evaluation method of renewable 30. Rani, P., Mishra, A.R.: Multi-criteria weighted aggregated sum
energy investment based on fuzzy analytic hierarchy process. Int. product assessment framework for fuel technology selection
J. Glob. Energy Issues 45(2), 153–165 (2023) using q-rung orthopair fuzzy sets. Sustain. Prod. Consum. 24,
10. Karamoozian, A., Wu, D., Luo, C.: Risk assessment of renewable 90–104 (2020)
energy projects using a novel hybrid fuzzy approach. Int. J. Green 31. Ali, J.: A q-rung orthopair fuzzy MARCOS method using novel
Energy (2023). https://doi.org/10.1080/15435075.2023.2166789 score function and its application to solid waste management.
11. Zadeh, L.A.: Fuzzy sets. Inform. Control 8(3), 338–353 (1965) Appl. Intell. 52(8), 8770–8792 (2022)
12. Atanassov, K.T.: More on intuitionistic fuzzy sets. Fuzzy Sets 32. Gomes, L.F.A.M.: An application of the TODIM method to the
Syst. 33, 37–46 (1989) multicriteria rental evaluation of residential properties. Eur.
13. Atanassov, K.T., Atanassov, K.T.: Interval valued intuitionistic J. Oper. Res. 193(1), 204–211 (2009)
fuzzy sets. In: Theory and Applications, Intuitionistic Fuzzy Sets, 33. Ren, P., Xu, Z., Gou, X.: Pythagorean fuzzy TODIM approach to
pp. 139–177 (1999) multi-criteria decision making. Appl. Soft Comput. 42, 246–259
14. Yager, R.R.: Pythagorean membership grades in multicriteria (2016)
decision making. IEEE Trans. Fuzzy Syst. 22(4), 958–965 (2013) 34. Zhang, Z., Guo, J., Zhang, H., Zhou, L., Wang, M.: Product
15. Peng, X., Yang, Y.: Fundamental properties of interval-valued selection based on sentiment analysis of online reviews: an
Pythagorean fuzzy aggregation operators. Int. J. Intell. Syst. intuitionistic fuzzy TODIM method. Complex Intell. Syst. 8(4),
31(5), 444–487 (2016) 3349–3362 (2022)
16. Yager, R.R.: Generalized orthopair fuzzy sets. IEEE Trans. Fuzzy 35. Liu, Y., Qin, Y., Liu, H.B., Xu, L.: Generalized q-ROF TODIM
Syst. 25(5), 1222–1230 (2017) method and its applications. Control Decis. (2020). https://doi.
17. Liu, P., Wang, P.: Some q-rung orthopair fuzzy aggregation org/10.13195/j.kzyjc.2018.1683
operators and their applications to multiple-attribute decision 36. Zhao, M., Gao, H., Wei, G., Wei, C., Guo, Y.: Model for network
making. Int. J. Intell. Syst. 33(2), 259–280 (2018) security service provider selection with probabilistic uncertain

123
International Journal of Fuzzy Systems

linguistic TODIM method based on prospect theory. Technol. its application in evaluation and selecting cold chain logistics
Econ. Dev. Econ. 28(3), 638–654 (2022) distribution center. Eng. Appl. Artif. Intell. 116, 105401 (2022)
37. He, S.S., Wang, Y.T., Peng, J.J., Wang, J.Q.: Risk ranking of 54. Ling, J., Lin, M., Zhang, L.: Medical waste treatment
wind turbine systems through an improved FMEA based on scheme selection based on single-valued neutrosophic numbers.
probabilistic linguistic information and the TODIM method. AIMS Math. 6(10), 10540–10564 (2021)
J. Oper. Res. Soc. 73(3), 467–480 (2022) 55. Lin, M., Wei, J., Xu, Z., Chen, R.: Multiattribute group decision-
38. Yang, G.F., Ren, M., Hao, X.M.: Multi-criteria decision-making making based on linguistic Pythagorean fuzzy interaction parti-
problem based on the novel probabilistic hesitant fuzzy entropy tioned Bonferroni mean aggregation operators. Complexity
and TODIM method. Alex. Eng. J. 68, 437–451 (2023) 2018(2), Article ID 9531064 (2018)
39. Sun, H., Yang, Z., Cai, Q., Wei, G.W., Mo, Z.W.: An extended 56. Zhou, X., Lin, M., Wang, W.: Statistical correlation coefficients
Exp-TODIM method for multiple attribute decision making based for single-valued neutrosophic sets and their applications in
on the Z-Wasserstein distance. Expert Syst. Appl. 214, 119114 medical diagnosis. AIMS Math. 8(7), 16340–16359 (2023)
(2024) 57. Rong, Y., Liu, Y., Pei, Z.: A novel multiple attribute decision-
40. Alali, F., Tolga, A.C.: Portfolio allocation with the TODIM making approach for evaluation of emergency management
method. Expert Syst. Appl. 124, 341–348 (2019) schemes under picture fuzzy environment. Int. J. Mach. Learn.
41. Alattas, K., Wu, Q.: A framework to evaluate the barriers for Cybern. 13(3), 633–661 (2022)
adopting the internet of medical things using the extended gen- 58. Liu, Y., Li, Y., Liang, H., Dong, Y.: Strategic experts’ weight
eralized TODIM method under the hesitant fuzzy environment. manipulation in 2-rank consensus reaching in group decision
Appl. Intell. 52(12), 13345–13363 (2022) making. Expert Syst. Appl. 216, 119432 (2023)
42. Nie, R.X., Tian, Z.P., Kwai Sang, C., Wang, J.Q.: Implementing 59. Yu, S., Du, Z., Zhang, X.: Clustering analysis and punishment-
healthcare service quality enhancement using a cloud-support driven consensus-reaching process for probabilistic linguistic
QFD model integrated with TODIM method and linguistic dis- large-group decision-making with application to car-sharing
tribution assessments. J. Oper. Res. Soc. 73(2), 207–229 (2022) platform selection. Int. Trans. Oper. Res. 29(3), 2002–2029
43. Ding, N., Yu, S.H., Chu, J.J., Chen, C., Shu, X.Y.: A decision (2022)
framework for cultural and creative products based on IF-TODIM 60. Qin, J., Liang, Y.: Consensus driven preference analysis model
method and group consensus reaching model. Adv. Eng. Inform. for incomplete group decision making with the multiple criteria
55, 101891 (2023) hierarchy process. Comput. Ind. Eng. (2023). https://doi.org/10.
44. Lance, G.N., Williams, W.T.: Computer programs for hierarchi- 1016/j.cie.2023.109024
cal polythetic classification (‘‘similarity analyses’’). Comput. J. 61. Liang, D., Li, F., Xu, Z.: A group-based FMEA approach with
9(1), 60–64 (1966) dynamic heterogeneous social network consensus reaching model
45. Hafezalkotob, A., Hafezalkotob, A.: A novel approach for com- for uncertain reliability assessment. J. Oper. Res. Soc. (2023).
bination of individual and group decisions based on fuzzy best- https://doi.org/10.1080/01605682.2021.2020694
worst method. Appl. Soft Comput. 59, 316–325 (2017)
46. Hafezalkotob, A., Hafezalkotob, A., Liao, H., Herrera, F.: Inter-
val MULTIMOORA method integrating interval Borda rule and Springer Nature or its licensor (e.g. a society or other partner) holds
interval best-worst-method-based weighting model: case study on exclusive rights to this article under a publishing agreement with the
hybrid vehicle engine selection. IEEE Trans. Cybern. 50(3), author(s) or other rightsholder(s); author self-archiving of the
1157–1169 (2019) accepted manuscript version of this article is solely governed by the
47. Tversky, A., Kahneman, D.: Advances in prospect theory: terms of such publishing agreement and applicable law.
cumulative representation of uncertainty. J. Risk Uncertain. 5,
297–323 (1992)
48. Rieger, M., Bui, T.: Too risk-averse for prospect theory? Mod. Yi Liu received the M.S.
Econ. 2(4), 691–700 (2011) degrees in Pure Mathematics
49. Hao, Z., Xu, Z., Zhao, H., Fujita, H.: A dynamic weight deter- from Sichuan Normal Univer-
mination approach based on the intuitionistic fuzzy Bayesian sity, Chengdu, China, in 2007,
network and its application to emergency decision making. IEEE and the Ph.D. degree in Com-
Trans. Fuzzy Syst. 26(4), 1893–1907 (2017) puter Sciences and Technology
50. Leoneti, A.B., Gomes, L.F.A.M.: A novel version of the TODIM from Southwest Jiaotong
method based on the exponential model of prospect theory: the University, Chengdu, China, in
ExpTODIM method. Eur. J. Oper. Res. 295(3), 1042–1055 2014. He is currently a professor
(2021) with the data recovery key lab-
51. Boran, F.E., Genç, S., Kurt, M., Akay, D.: A multi-criteria oratory of Sichuan Province and
intuitionistic fuzzy group decision making for supplier selection School of Mathematics and
with TOPSIS method. Expert Syst. Appl. 36(8), 11363–11368 Information Sciences, Neijiang
(2009) Normal University, Sichuan,
52. Senapati, T., Simic, V., Saha, A., Dobrodolac, M., Rong, Y., China. He has authored or
Tirkolaee, E.B.: Intuitionistic fuzzy power Aczel-Alsina model coauthored more than 60 publications. His research interests include
for prioritization of sustainable transportation sharing practices. aggregation operators, fuzzy decision making, automated reasoning,
Eng. Appl. Artif. Intell. 119, 105716 (2023) and their applications.
53. Rong, Y., Yu, L., Niu, W., Liu, Y., Senapati, T., Mishra, A.R.:
MARCOS approach based upon cubic Fermatean fuzzy set and

123
Y. Liu et al.: Prospect Theory-Based q-Rung...

Ya Qin received the M.S. Abdul Wali Khan University, Mardan, Pakistan. The research areas of
degrees in Applied Mathematics Dr. Saleem are fuzzy logic, fuzzy set theory, Fuzzy Aggregation
from Sichuan Normal Univer- operators and decision making. He has published more than 330
sity, Chengdu, China, in 2009. research articles in international peer-reviewed journals, including ISI
She is currently an associated Indexed/IF Journal publications with cumulative impact factor (I.F) of
Professor with the School of the published papers more than 900. He can be considered as one of
Mathematics and Information the best research fellows at the Department of Mathematics Abdul
Sciences, Neijiang Normal Wali Khan University Mardan. Some papers have been published in
University, Sichuan, China. She high-impact journals including International Journal of Intelligent
has authored or coauthored Systems, Soft Computing, International Journal of General Systems,
more than 15 publications. Her Journal of Intelligent and Fuzzy Systems, International Journal of
research interests include Fuzzy Systems, Journal of Ambient Intelligence and Humanized
aggregation operators, fuzzy Computing, and Symmetry. He has supervised 12 Ph.D. scholars and
decision making and their 30 MS/M.Phil. Students. He has engaged as editorial board member
applications. in 5 Journals and reviewer of more than 70 Journals. He also pub-
lished 10 articles in conference proceeding in international confer-
Haobin Liu received the M.S. ence. He delivered more than 30 Talks in international conferences in
degrees in computational Math- Pakistan and abroad. He also visited UAE, Oman, Saudi Arabia,
ematics from Chongqing Turkey, Russia, China, Thailand, Sri Lanka. He is a very active
University, Chongqing, China, researcher and he also collaborated more than 50 researchers in dif-
in 2010. He is currently a Lec- ferent countries.
ture with the School of Mathe-
matics and Information Yuan Rong received a Master
Sciences, Neijiang Normal of Science degree in Mathe-
University, Sichuan, China. He matics and Applied Mathemat-
presided over three projects, ics from Xihua University in
guided students to participate in 2021. He is studying for a doc-
academic competitions, and torate in management science
won 14 international awards. He and engineering at Shanghai
has authored or coauthored University, China. He has
more than 10 publications. His authored or coauthored more
research interests include aggregation operators, fuzzy decision than 20 publications. His current
making and their applications. research interests include
uncertainty decision analysis,
Saleem Abdullah was born in computing with words, decision
the Village of Yaqubi, Swabi, support systems, risk manage-
Pakistan in 1986. Dr. Saleem ment and so forth.
completed the early schooling
from Government High School
Yar Hussain and College degree
from Government Postgraduate
College Mardan. He has com-
pleted MSc in Mathematics
from Hazara University Man-
sehra. He received his MPhil
and Ph.D. degrees in Mathe-
matics from Quaid-i-Azim
University, Islamabad in 2011
and 2015, respectively. He is an
Assistant Professor of Mathematics at Department of Mathematics,

123

You might also like