research article a multiple attribute group decision making ...a procedure that uses a fuzzy...
TRANSCRIPT
-
Research ArticleA Multiple Attribute Group Decision Making Approach forSolving Problems with the Assessment of Preference Relations
Taho Yang,1 Yiyo Kuo,2 David Parker,3 and Kuan Hung Chen1
1 Institute of Manufacturing Information and Systems, National Cheng Kung University, Tainan 70101, Taiwan2Department of Industrial Engineering and Management, Ming Chi University of Technology, New Taipei City 24301, Taiwan3The University of Queensland Business School, Brisbane, QLD 4072, Australia
Correspondence should be addressed to Yiyo Kuo; [email protected]
Received 19 June 2014; Revised 21 October 2014; Accepted 23 October 2014
Academic Editor: Mu-Chen Chen
Copyright ยฉ 2015 Taho Yang et al. This is an open access article distributed under the Creative Commons Attribution License,which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
A number of theoretical approaches to preference relations are used for multiple attribute decision making (MADM) problems,and fuzzy preference relations is one of them. When more than one person is interested in the same MADM problem, it thenbecomes a multiple attribute group decision making (MAGDM) problem. For both MADM and MAGDM problems, consistencyamong the preference relations is very important to the result of the final decision. The research reported in this paper is based ona procedure that uses a fuzzy preference relations matrix which satisfies additive consistency. This matrix is used to solve multipleattribute group decision making problems. In group decision problems, the assessment provided by different experts may divergeconsiderably. Therefore, the proposed procedure also takes a heterogeneous group of experts into consideration. Moreover, themethods used to construct the decision matrix and determine the attribution of weight are both introduced. Finally a numericalexample is used to test the proposed approach; and the results illustrate that the method is simple, effective, and practical.
1. Introduction
There are many situations in daily life and in the workplacewhich pose a decision problem. Some of them involve pickingthe optimum solution from amongmultiple available alterna-tives. Therefore in many domain problems multiple attributedecision making methods, such as simple additive weighting(SAW), the technique for order preference by similarity toideal solution (TOPSIS), analytical hierarchy process (AHP),data envelopment analysis (DEA), or grey relational analysis(GRA) [1โ5], are usually adopted, for example, layout design[6โ8], supply chain design [9], push/pull junction pointselection [10], pacemaker location determination [11], workin process level determination [12], and so on.
If more than one person is involved in the decision, thedecision problem becomes a group decision problem. Manyorganizations have moved from a single decision maker orexpert to a group of experts (e.g., Delphi) to accomplish thistask successfully [13, 14]. Note that an โexpertโ represents an
authorized person or an expert who should be involved inthis decision making process. However, no single alternativeworks best for all performance attributes, and the assessmentof each alternative given by different decision makers maydiverge considerably. As a consequence, multiple attributegroup decision making (MAGDM) is more difficult thancases where a single decision maker decides using a multipleattribute decision making method.
MAGDMis one of themost common activities inmodernsociety, which involves selecting the optimal one from afinite set of alternatives with respect to a collection ofthe predefined criteria by a group of experts with a highcollective knowledge level on these particular criteria [15].When a group of experts wants to choose a solution fromamong several alternatives, preference relations is one typeof assessment that experts could provide. Preference relationsare comparisons between two alternatives for a particularattribute. A higher preference relation means that there is ahigher degree of preference for one alternative over another.
Hindawi Publishing CorporationMathematical Problems in EngineeringVolume 2015, Article ID 849897, 10 pageshttp://dx.doi.org/10.1155/2015/849897
-
2 Mathematical Problems in Engineering
However, different expertsmay use different assessment typesto express the preference relation. It is possible that in groupdecision making different experts express their preference indifferent formats [16โ21].
In addition, after experts have provided their assessmentof the preference relation, the appropriateness of the compar-ison from each expert must be tested. Consistency is one ofthe important properties for verifying the appropriateness ofchoices [22]. If the comparison from an expert is not logicallyconsistent for a specific attribute, it means that at leastone preference relation provided by the expert is defective.Therefore, the lack of consistency in decisionmaking can leadto inconsistent conclusions.
Quite apart from the type of assessment, there can beconsiderable variation between experts as to their evaluationof the degree of the preference relation. In general, it would bepossible to aggregate the preferences of experts by taking theweight assigned by every expert into consideration. However,heterogeneity among experts should also be considered [23].For example, if the expert who assigns the greatest weightto a preference relation also makes choices that are notappropriate and quite different from the evaluations of theother experts who assign lower weights, then the groupdecision procedure can be distorted and imperfect.
Moreover, the determination of attribute weight is also animportant issue [24]. In some decision cases, some attributesare considered to be more important in the expertsโ profes-sional judgment. However, for these important attributes, thepreference relation provided by experts may be quite similarfor all alternatives. Even for the attribute with the highestweight, the degree of influence on the final decision wouldbe very small in this case. In this way this kind of attributecan become unimportant to the final decision [25].
Therefore, during the multiple attribute group decisionprocess, 5 aspects should be noted:
(i) considering different assessment types simultane-ously;
(ii) insuring the preference relations provided by expertsare consistent;
(iii) taking heterogeneous experts into consideration;(iv) deciding the weight of each attribute;(v) ranking all alternatives.
Group decision making has been addressed in the lit-erature. In recent years, Oฬlcฬงer and Odabasฬงi [23] proposeda fuzzy multiple attribute decision making method to dealwith the problem of ranking and selecting alternatives.Experts provide their opinion in the form of a trapezoidalfuzzy number. These trapezoidal fuzzy numbers are thenaggregated and defuzzified into a MADM. Finally, TOPSISis used to rank and select alternatives. In the method, expertscan provide their opinion only by trapezoidal fuzzy number.
Boran et al. [26] proposed a TOPSIS method combinedwith intuitionistic fuzzy set to select appropriate supplierin group decision making environment. Intuitionistic fuzzyweighted averaging (IFWA) operator is utilized to aggre-gate individual opinions of decision makers for rating the
importance of criteria and alternatives. Cabrerizo et al. [27]presented a consensus model for group decision makingproblems with unbalanced fuzzy linguistic information. Thisconsensus model is based on both a fuzzy linguistic method-ology to deal with unbalanced linguistic term sets and twoconsensus criteriaโconsensus degrees and proximity mea-sures. Chuu [28] builds a group decisionmakingmodel usingfuzzy multiple attributes analysis to evaluate the suitability ofmanufacturing technology. The proposed approach involveddeveloping a fusion method of fuzzy information, which wasassessed using both linguistic and numerical scales.
Lu et al. [29] developed a software tool for support-ing multicriteria group decision making. When using thesoftware, after inputting all criteria and their correspondingweights, and the weighting for all the experts, all the expertscan assess every alternative against each attribute. Then theranking of all alternatives can be generated. In the softwareonly one assessment type is allowed and there is no functionthat can be used to ensure that the preference relationsprovided by experts are consistent. Zhang and Chu [30]proposed a group decision making approach incorporatingtwo optimization models to aggregate these multiformat andmultigranularity linguistic judgments. Fuzzy set theory isutilized to address the uncertainty in the decision makingprocess.
Cabrerizo et al. [14] proposed a consensus model to dealwith group decision making problems, in which experts useincomplete unbalanced fuzzy linguistic preference relationsto provide their preference. However, the model requiresthat preference relations should be assessed in the sameway, and no allowance is made for heterogeneous experts.Cebi and Kahraman [31] proposed a methodology for groupdecision support. The methodology consists of eight stepswhich are (1) definition of potential decision criteria, possiblealternatives, and experts, (2) determining the weighting ofexperts, (3) identifying the importance of criteria, (4) assign-ing alternatives, (5) aggregating expertsโ preferences, (6)identifying functional requirements, (7) calculating informa-tion contents, and (8) calculating weighted total informationcontents and selecting the best alternative. The methodologydoes not include a check on the consistency of preferencerelations provided by the experts.
The novelty of the present study is that it proposes amultiple attribute group decision making methodology inwhich all of the five issues mentioned above are addressed.A review of the literature related to this research suggeststhat no previous research has addressed all of the issuessimultaneously. For managers who are not experts in fuzzytheory, group decision making, MADM, and so on, thisresearch can provide a complete guideline for solving theirmultiple attribute group decision making problem.
The remainder of this paper is organized as follows.In Section 2 all the issues set out above are discussed andappropriate methodologies for dealing with them are pro-posed. Then an overall approach is proposed in Section 3.The proposedmodel is tested and examined with a numericalexample in Section 4. Finally Section 5 contains the discus-sion and conclusions.
-
Mathematical Problems in Engineering 3
2. Multiple Attribute GroupDecision Making Methodology
2.1. Assessment and Transformation of Preference Relations.There are two types of preference relations that are widelyused. One is fuzzy preference relations, in which ๐
๐๐denotes
the preference degree or intensity of the alternative ๐ over ๐[32โ35]. If ๐
๐๐= 0.5, it means that alternatives ๐ and ๐ are
indifferent; if ๐๐๐= 1, it means that alternative ๐ is absolutely
preferred to ๐, and if ๐๐๐> 0.5, it means that alternative ๐ is
preferred to ๐. ๐๐๐is reciprocally additive; that is, ๐
๐๐+ ๐๐๐= 1
and ๐๐๐= 0.5 [35, 36].
The other widely used type of preference relations is mul-tiplicative preference relations, in which ๐
๐๐indicates a ratio
of preference intensity for alternative ๐ to that of alternative ๐;that is, it is interpreted asmeaning that alternative ๐ is ๐
๐๐times
as good as alternative ๐ [17]. Saaty [3] suggested measuring๐๐๐on an integer scale ranging from 1 to 9. If ๐
๐๐= 1, it
means that alternatives ๐ and ๐ are indifferent; if ๐๐๐= 9, it
means that alternative ๐ is absolutely preferred to ๐, and if8 โฅ ๐
๐๐โฅ 2, it means that alternative ๐ is preferred to ๐. In
addition, ๐๐๐ร ๐๐๐= 1, and ๐
๐๐= ๐๐๐ร ๐๐๐.
For these two preference types, Chiclana et al. [17] pro-posed an equation to transform the multiplicative preferencerelation into the fuzzy preference relation, as shown by
๐๐๐= 0.5 (1 + log
9๐๐๐) . (1)
However, for both preference types, it is possible thatsome experts would not wish to provide their preferencerelation in the form of a precise value. In the fuzzy preferencerelations, experts can use the following classifications:
(i) a precise value, for example, โ0.7โ;(ii) a range, for example, (0.3, 0.7); the value is likely to
fall between 0.3 and 0.7;(iii) a fuzzy number with triangular membership func-
tion, for example, (0.4, 0.6, 0.8); the value is between0.4 and 0.8 and is most probably 0.6;
(iv) a fuzzy number with trapezoidal membership func-tion, for example, (0.3, 0.5, 0.6, 0.8); the value isbetween 0.3 and 0.8, most probably between 0.5 and0.6.
In this paper, the four classifications set out above areunified by transferring them into trapezoidal membershipfunctions. Thus, 0.7 becomes (0.7, 0.7, 0.7, 0.7), (0.3, 0.7)becomes (0.3, 0.3, 0.7, 0.7), and (0.4, 0.6, 0.8) then becomes(0.4, 0.6, 0.6, 0.8). If experts provide their assessment inthe format of multiplicative preference relations, it will betransformed into a trapezoidal membership function first,and then using (1) it will be further transformed into theformat of fuzzy preference relations. For example, (3, 4, 5,6) can be transferred into (0.75, 0.82, 0.87, 0.91) by using(1). Therefore, this paper will mention only fuzzy preferencerelations in what follows.
2.2. The Generation of Consistent Preference Relations. Theproperty of consistency has been widely used to establish
a verification procedure for preference relations, and it isvery important for designing good decision making models[22]. In the analytical hierarchy process, for example, inorder to avoid potential comparative inconsistency betweenpairs of categories, a consistency ratio (CR), an index forconsistency, has been calculated to assure the appropriatenessof the comparisons [3]. If the CR is small enough, there isno evidence of inconsistency. However, if the CR is too high,then the experts should adjust their assessments again andagain until the CR decreases to a reasonable value. For fuzzypreference relations, Herrera-Viedma et al. [22] designeda method for constructing consistent preference relationswhich satisfy additive consistency. Using this method, allexperts need only to provide preference relations betweenalternatives ๐ and ๐ + 1, ๐
๐(๐+1), and the remaining preference
relations can be calculated using (2) if ๐ > ๐ and (3) if ๐ < ๐:
๐๐๐=
๐ โ ๐ + 1
2โ ๐๐(๐+1)
โ ๐(๐+1)(๐+2)
โ โ โ โ โ ๐(๐โ1)๐
, โ๐ > ๐,
(2)
๐๐๐= 1 โ ๐
๐๐, โ๐ < ๐. (3)
To illustrate the generation of preferential relations, weprovide an empirical example of four alternatives as follows.First, the expert provides the three preference relations as๐12
= 0.3, ๐23
= 0.6, and ๐34
= 0.8.According to (2),
๐21
= 1 โ 0.3 = 0.7,
๐31
= 1.5 โ 0.3 โ 0.6 = 0.6,
๐41
= 2 โ 0.3 โ 0.6 โ 0.8 = 0.3,
๐32
= 1 โ 0.6 = 0.4,
๐42
= 1.5 โ 0.6 โ 0.8 = 0.1,
๐43
= 1 โ 0.8 = 0.2.
(4)
According to (3),
๐13
= 1 โ 0.6 = 0.4,
๐14
= 1 โ 0.3 = 0.7,
๐24
= 1 โ 0.1 = 0.9.
(5)
Therefore, the preference relations matrix, PR, is
PR =[[[
[
0.5 0.3 0.4 0.7
0.7 0.5 0.6 0.9
0.6 0.4 0.5 0.8
0.3 0.1 0.2 0.5
]]]
]
. (6)
In general, experts are asked to evaluate all pairs ofalternatives and then construct a preference matrix with fullinformation. However, it is difficult to obtain a consistentpreference matrix in practice, especially when measuringpreferences on a set with a large number of alternatives [22].
-
4 Mathematical Problems in Engineering
2.3. Assessment Aggregation for a Heterogeneous Group ofExperts. For each comparison between a pair of alternatives,the preference relations provided by different experts wouldvary. Hsu and Chen [37] proposed an approach to aggregatefuzzy opinions for a heterogeneous group of experts. Then,Chen [38]modified the approach and Oฬlcฬงer andOdabasฬงi [23]present it as the following six-step procedure.
(1) Calculate the Degree of Agreement between Each Pairof Experts. For a comparison between two alternatives, letthere be ๐ธ experts in the decision group, (๐
1, ๐2, ๐3, ๐4) and
(๐1, ๐2, ๐3, ๐4) are the preference relations provided by experts
๐ and ๐, 1 โค ๐ โค ๐ธ, 1 โค ๐ โค ๐ธ, and ๐ ฬธ= ๐. The similaritybetween these two trapezoidal fuzzy numbers, ๐
๐๐, can be
measured by
๐๐๐
= 1 โ
๐1 โ ๐1 +
๐2 โ ๐2 +
๐3 โ ๐3 +
๐4 โ ๐4
4. (7)
(2) Construct the Agreement Matrix. After all the agreementdegrees between experts are measured, the agreement matrix(AM) can be constructed as follows:
AM =[[[[
[
1 ๐12
โ โ โ ๐1๐ธ
๐21
1 โ โ โ ๐2๐ธ
...... ๐๐๐
...
๐๐ธ1
๐๐ธ2
โ โ โ 1
]]]]
]
, (8)
in which ๐๐๐
= ๐๐๐, and if ๐ = ๐, then ๐
๐๐= 1.
(3) Calculate the AverageDegree of Agreement for Each Expert.The average degree of agreement for expert ๐ (AA
๐) can be
calculated by
AA๐=
1
๐ธ โ 1
๐ธ
โ๐=1,๐ ฬธ=๐
๐๐๐, โ๐. (9)
(4) Calculate the RelativeDegree of Agreement for Each Expert.After calculating the average degree of agreement for allexperts, the relative degree of agreement for expert ๐ (RA
๐)
can be calculated by
RA๐=
AA๐
โ๐ธ
๐=1AA๐
, โ๐. (10)
(5) Calculate the Coefficient for the Degree of Consensusfor Each Expert. Let ew
๐be the weight of expert ๐, and
โ๐ธ
๐=1ew๐= 1. The coefficient of the degree of consensus for
expert ๐ (CC๐) can be calculated by
CC๐= ๐ฝ โ ew
๐+ (1 โ ๐ฝ) โ RA
๐, โ๐, (11)
in which ๐ฝ is a relaxation factor of the proposed method and0 โค ๐ฝ โค 1. It represents the importance of ew
๐over RA
๐.
When ๐ฝ = 0, it means that the group of experts is consideredto be homogeneous.
(6) Calculate the Aggregation Result. Finally, the aggregationresult of the comparison between two alternatives ๐ and ๐ is๐๐๐, where
๐๐๐= CC
1โ ๐๐๐ (1) โ CC2 โ ๐๐๐ (2) โ โ โ โ โ CC๐
โ ๐๐๐ (๐) โ โ โ โ โ CC๐ธ โ ๐๐๐ (๐ธ) .
(12)
In (12), ๐๐๐(๐) is the preference relation between alterna-
tives ๐ and ๐ provided by expert ๐, and ๐๐๐= (๐1๐๐, ๐2๐๐, ๐3๐๐, ๐4๐๐).
Moreover โ and โ are the fuzzy multiplication operator andthe fuzzy addition operator, respectively.
Let there be ๐ alternatives. Since each expert onlyprovides preference relations between alternatives ๐ and ๐ +1, the aggregation process for a heterogeneous group ofexperts must be executed ๐ โ 1 times in order to generate๐ โ 1 aggregated trapezoidal fuzzy numbers. These ๐ โ1 trapezoidal fuzzy numbers can then be converted into aprecise value by the use of
๐๐๐=๐1๐๐+ 2 (๐2
๐๐+ ๐3๐๐) + ๐4๐๐
6. (13)
After the aggregation procedure, using (2) and (3), anaggregated preference relations matrix for attribute ๐ isconstructed as follows:
PR๐=
[[[[
[
1 ๐12
โ โ โ ๐1๐
๐12
1 โ โ โ ๐2๐
...... 1
...
๐๐1
๐๐2
โ โ โ 1
]]]]
]
. (14)
2.4. AttributeWeightDetermination. In a preference relationsmatrix of attribute ๐, ๐
๐๐indicates the degree of preference
of alternative ๐ over ๐ when attribute ๐ was considered.Therefore, โ๐
๐=1,๐ ฬธ=๐๐๐๐indicates total degree of preference of
alternative ๐ over the other ๐ โ 1 alternatives. In the sameway, โ๐
๐=1,๐ ฬธ=๐๐๐๐indicates the total degree of preference of the
other๐โ1 alternatives over alternative ๐. Fodor and Roubens[39] proposed (15) to define ๐ฟ
๐๐, the net degree of preference
of alternative ๐ over the other ๐ โ 1 alternatives by attribute๐, and the bigger ๐ฟ
๐๐is, the better alternative ๐ by attribute ๐
is:
๐ฟ๐๐=
๐
โ๐=1,๐ ฬธ=๐
๐๐๐โ
๐
โ๐=1,๐ ฬธ=๐
๐๐๐, โ๐, ๐. (15)
Thus, the problem is reduced to a multiple attributedecision making problem:
DM =[[[[
[
๐ฟ11
๐ฟ12
โ โ โ ๐ฟ1๐
๐ฟ21
๐ฟ22
โ โ โ ๐ฟ2๐
......
......
๐ฟ๐1
๐ฟ๐2
โ โ โ ๐ฟ๐๐
]]]]
]
. (16)
-
Mathematical Problems in Engineering 5
For the decision matrix constructed in Section 2.4, Wangand Fan [25] proposed two approaches, absolute deviationmaximization (ADM) and standard deviation maximization(SDM), to determine the weight of all attributes. For a certainattribute, if the difference of the net degree of preferenceamong all alternatives shows a wide variation, this means thisattribute is quite important. ADM and SDM used absolutedeviation (AD) and standard deviation (SD) to measure thedegree of variation. An attribute with a bigger value of ADand SD will be a more important attribute.
When ADM was adopted, the weight of attribute ๐, aw๐,
was calculated by using (17), while if SDM was adopted, (18)was used for calculating the weight of attribute ๐:
aw๐=
(โ๐
๐=1โ๐
๐=1
๐ฟ๐๐โ ๐ฟ๐๐
)1/(๐โ1)
โ๐
๐=1(โ๐
๐=1โ๐
๐=1
๐ฟ๐๐โ ๐ฟ๐๐
)1/(๐โ1)
, โ๐; ๐ > 1, (17)
aw๐=
(โ๐
๐=1๐ฟ2๐๐)1/2(๐โ1)
โ๐
๐=1(โ๐
๐=1๐ฟ2๐๐)1/2(๐โ1)
, โ๐; ๐ > 1, (18)
where ๐ is the parameter of these two functions for calcu-lating weights. Setting the variable to different values willlead to different weights and when ๐ = โ, all weightswill be equal. Therefore, in order to reflect the differencesamong the attribute weights, Wang and Fang [25] suggestedpreferring a small value for parameter ๐. Further details ofthe demonstration of the use of ADM and SDM can be foundin the paper by Wang and Fan [25].
2.5. Alternative Ranking. Once the weights of all attributesare determined by (17) or (18), the multiple attribute decisionmaking problem constructed by (16) can be solved by theapplication of a multiple attribute decision making method,such as SAW, TOPSIS, ELECTRE, or GRA [1, 2, 5]. Accordingto Kuo et al. [40], different MADM methods would lead todifferent results, but similar ranking of alternatives. In thisresearch, SAW was selected for the MADM problem. Sincethe weight calculated by (17) and (18) has been normalized,and โ๐
๐=1aw๐
= 1, the score of alternatives ๐, ๐ถ๐, can be
calculated directly by
๐ถ๐=
๐
โ๐=1
aw๐๐ฟ๐๐, ๐ = 1, 2, . . . , ๐. (19)
The bigger the๐ถ๐is, the better the alternative ๐ is. After the
scores of all alternatives have been calculated, the alternativescan be ranked by ๐ถ
๐.
3. The Proposed Approach
Following from the consideration of issues whichwere set outin the Introduction and further developed in Section 2, thisresearch proposes a 5-step procedure for multiple attributegroup decision making problems as shown in Figure 1.
In Step 1, experts provide their preference relations forall attributes using their preferred format of expression. In
transformation
heterogeneous group of experts
relations
(1) Preference relations assessment and
(2) Assessment aggregation for
(3) The generation of consistent preference
(4) Attribute weight determination
(5) Alternatives ranking
Figure 1: The proposed MAGDM procedure.
order to ensure the additive consistency of these preferencerelations, only the preference relations between alternatives ๐and ๐+1 are assessed.Then these preference relations providedby the experts are transformed into trapezoidal membershipfunctions. If the preference relations are multiplicative pref-erence relations, (1) is used to transform them into fuzzypreference relations.
In Step 2, in order to take the heterogeneity of the expertsinto consideration, the trapezoidal membership function offuzzy preference relations for all experts is aggregated by a six-step procedure given by Oฬlcฬงer andOdabasฬงi [23].Then (2) and(3) are used to calculate the remaining preference relationswhich had not been provided by the experts, and these arethen used to construct preference relationmatrixes which areadditively consistent in Step 3.
In Step 4, these preference relation matrixes are trans-formed into a traditional multiple attribute decision matrixand used to determine the weight of all attributes using (17)and (18). Finally, all the scores of alternatives can be calculatedusing (19) and the alternatives can be ranked in Step 5.
4. Numerical Example
The proposed MAGDM methodology allows two types ofpreference relations, fuzzy reference relations andmultiplica-tive preference relations, which are explained in Section 2.1.The former ones are transformed to numerical numberthrough fuzzy membership functions and the latter onesdirectly use numerical numbers. They are then aggregatedthrough the proposed aggregation and ranking procedure asdiscussed in Sections 2.2 to 2.5. Due to both the transforma-tion and aggregation procedures, the resulting numbers arereal numbers.
-
6 Mathematical Problems in Engineering
In this section, we provide a numerical example toillustrate the implementation of the proposed methodology.Consider four alternatives, three experts, and two attributeMAGDM problems as follows.
Step 1 (preference relations assessment and transformation).The preference relations assessments of Attribute 1 providedby these three experts were given as follows. in which ๐
๐๐is
the assessment of attribute ๐ provided by expert ๐:
๐ 11
=[[[
[
โ Low โ โโ โ Low โโ โ โ Mediumโ โ โ โ
]]]
]
,
๐ 21
=[[[
[
โ More low โ โโ โ Medium โโ โ โ Mediumโ โ โ โ
]]]
]
,
๐ 31
=
[[[[[[
[
โ1
3โ โ
โ โ1
4โ
โ โ โ 1
โ โ โ โ
]]]]]]
]
.
(20)
In this example, Experts 1 and 2 preferred to provideassessment by fuzzy preference relations, and Expert 3 pre-ferred to provide assessment by multiplicative preferencerelations. However, Expert 1 used the membership functionas shown in Figure 2, Expert 2 used themembership functionas shown in Figure 3, and Expert 3 used precise values forproviding his/her preference relations. All assessments arethen transformed into the type of trapezoidal membershipfunction as shown below:
๐ 11
=[[[
[
โ 0.125, 0.225, 0.325 0.425 โ โ
โ โ 0.125, 0.225, 0.325, 0.425 โ
โ โ โ 0.350, 0.450, 0.550, 0.650
โ โ โ โ
]]]
]
,
๐ 21
=[[[
[
โ 0.200, 0.300, 0.400, 0.500 โ โ
โ โ 0.350, 0.450, 0.550, 0.650 โ
โ โ โ 0.350, 0.450, 0.550, 0.650
โ โ โ โ
]]]
]
,
๐ 31
=
[[[[[[
[
โ1
3,1
3,1
3,1
3โ โ
โ โ1
4,1
4,1
4,1
4โ
โ โ โ 1, 1, 1, 1,
โ โ โ โ
]]]]]]
]
.
(21)
The preference relationsโ assessments of Attribute 2 whichhave been transformed into the type of trapezoidal member-ship function were given as follows:
๐ 12
=[[[
[
โ 0.125, 0.225, 0.325 0.425 โ โ
โ โ 0.350, 0.450, 0.550, 0.650 โ
โ โ โ 0.125, 0.225, 0.325, 0.425
โ โ โ โ
]]]
]
,
๐ 22
=[[[
[
โ 0.050, 0.150, 0.250, 0.350 โ โ
โ โ 0.500, 0.600, 0.700, 0.800 โ
โ โ โ 0.200, 0.300, 0.400, 0.500
โ โ โ โ
]]]
]
,
-
Mathematical Problems in Engineering 7
๐ 32
=
[[[[[[
[
โ1
4,1
4,1
4,1
4โ โ
โ โ 1, 1, 1, 1 โ
โ โ โ1
3,1
3,1
3,1
3,
โ โ โ โ
]]]]]]
]
.
(22)
Using (1) themultiplicative preference relations in๐ 31and
๐ 32
can be transformed into fuzzy preference relations and
then become ๐ 31
and ๐ 32
as follows. ๐ 31
and ๐ 32
were thenreplaced by ๐
31and ๐
32for the rest of the analysis:
๐
31=
[[[
[
โ 0.250, 0.250, 0.250, 0.250 โ โ
โ โ 0.185, 0.185, 0.185, 0.185 โ
โ โ โ 0.500, 0.500, 0.500, 0.500
โ โ โ โ
]]]
]
,
๐
32=
[[[
[
โ 0.185, 0.185, 0.185, 0.185 โ โ
โ โ 0.500, 0.500, 0.500, 0.500 โ
โ โ โ 0.250, 0.250, 0.250, 0.250
โ โ โ โ
]]]
]
.
(23)
Step 2 (assessment aggregation for heterogeneous group ofexperts). In this example, the weights of Experts 1, 2, and 3are 0.3, 0.3, and 0.4, respectively. Following the method setout in Section 2.3, the six steps can be used to aggregate theassessments provided by the heterogeneous group of experts.Let the relaxation factor ๐ฝ = 0.5. The results are thensummarized in Table 1.
Therefore, the aggregated preference relations matrixesPR1and PR
2are as shown in the following:
PR1=
[[[
[
โ 0.290 โ โ
โ โ 0.311 โ
โ โ โ 0.500
โ โ โ โ
]]]
]
,
PR2=
[[[
[
โ 0.218 โ โ
โ โ 0.547 โ
โ โ โ 0.290
โ โ โ โ
]]]
]
.
(24)
Step 3 (the generation of consistent preference relations). InStep 3, the results in PR
1and PR
2are incomplete. Equations
(2) and (3) are then used to calculate the remaining preferencerelations and to construct additively consistent preference
relation matrixes. The complete preference relation matrixesPR1and PR
2are
PR1=
[[[
[
0.500 0.290 0.100 0.100
0.710 0.500 0.311 0.311
0.900 0.689 0.500 0.500
0.900 0.689 0.500 0.500
]]]
]
,
PR2=
[[[
[
0.500 0.218 0.265 0.055
0.782 0.500 0.547 0.337
0.735 0.453 0.500 0.290
0.945 0.663 0.710 0.500
]]]
]
.
(25)
According to the proposition and proof from Herrera-Viedma et al. [22], a fuzzy preference relation PR = (๐
๐๐) is
consistent if and only if ๐๐๐+ ๐๐๐+ ๐๐๐= 3/2, โ๐ โค ๐ โค ๐. It can
be found that above PR1and PR
2are consistent.
Step 4 (attribute weight determination). Using (15) to calcu-late all ๐ฟ
๐๐, the decision matrix DM can be constructed as
follows:
DM =[[[
[
โ2.019 โ1.923
โ0.336 0.331
1.178 โ0.045
1.178 1.637
]]]
]
. (26)
According to the constructed decision matrix, whenADM and SDM were adopted, the weight of Attributes 1 and2 can be calculated by (17) and (18), respectively. A valueof ๐ = 2 has been adopted arbitrarily for the sake of thisdemonstration. If ADM is adopted, the weights of Attributes 1and 2 are 0.501 and 0.499, respectively. If SDM is adopted theweights of Attributes 1 and 2 are 0.509 and 0.491,respectively.
-
8 Mathematical Problems in Engineering
Table 1: Aggregation of heterogeneous group of experts for Attribute 1.
๐12
๐23
๐34
Expert 1 (0.125, 0.225, 0.325, 0.425) (0.125, 0.225, 0.325, 0.425) (0.350, 0.450, 0.550, 0.650)Expert 2 (0.200, 0.300, 0.400, 0.500) (0.350, 0.450, 0.550, 0.650) (0.350, 0.450, 0.550, 0.650)Expert 3 (0.250, 0.250, 0.250, 0.250) (0.185, 0.185, 0.185, 0.185) (0.500, 0.500, 0.500, 0.500)Degree of agreement (๐
๐๐)
๐12
0.925 0.775 1.000๐13
0.900 0.880 0.900๐23
0.875 0.685 0.900Average degree of agreement of expert ๐ (AA
๐)
AA1 0.913 0.828 0.950AA2 0.900 0.730 0.950AA3 0.888 0.783 0.900
Relative degree of agreement of expert ๐ (RA๐)
RA1 0.338 0.354 0.339RA2 0.333 0.312 0.339RA3 0.329 0.334 0.321
Consensus degree coefficient of expert ๐ (CC๐) for
๐ฝ = 0.5
CC1 0.319 0.327 0.320CC2 0.317 0.306 0.320CC3 0.364 0.367 0.361
Aggregated results ๐12 = (0.19, 0.26, 0.32, 0.38) ๐23 = (0.22, 0.28, 0.34, 0.41) ๐34 = (0.40, 0.47, 0.53, 0.60)Converted results ๐12 = 0.290 ๐23 = 0.311 ๐34 = 0.500
Fuzzy preference relation
Very highHighMediumLow
Very low
0.2 0.4 0.6 1.00.8
0.2
0.4
0.6
0.8
1.0
Mem
bers
hip
valu
e
Figure 2: Membership functions adopted by Expert 1.
Very highHighMediumLow
Very low
More highMore low
0.2
0.4
0.6
0.8
1.0
Mem
bers
hip
valu
e
Fuzzy preference relation 0.2 0.4 0.6 1.00.8
Figure 3: Membership functions adopted by Expert 2.
Table 2:The scoring results byweight determinationmethodsADMand SDM.
Alternative ๐ ๐ฟ๐๐
๐ถ๐(ADM) ๐ถ
๐(SDM) Ranking results
1 โ2.019 โ1.923 โ1.971 โ1.972 42 โ0.336 0.331 โ0.003 โ0.009 33 1.178 โ0.045 0.567 0.577 24 1.178 1.637 1.407 1.403 1aw๐by ADM 0.501 0.499
aw๐by SDM 0.509 0.491
Step 5 (ranking alternatives). After generating the weights ofAttributes 1 and 2, using SAW, the score of all alternatives๐ถ๐can be calculated by (9). The scoring results are as shown
in Table 2. In Table 2, ๐ถ๐(ADM) and ๐ถ
๐(SDM) indicate the
scores of all alternatives using attribute weight determiningapproaches ADM and SDM, respectively. The bigger valuesof ๐ถ๐indicate that the alternative ๐ is better. In the case of
the values of ๐ถ๐(ADM), for example, because ๐ถ
4(ADM)
> ๐ถ3(ADM) > ๐ถ
2(ADM) > ๐ถ
1(ADM), the group
decision selected Alternative 4 as the first priority. Moreover,according to the values of ๐ถ
๐(SDM), the results also show
Alternative 4 as the first priority.Although the theoretical development involves com-
plicated technical details, the implementation is relativelystraightforward in light of the numerical implementation.
-
Mathematical Problems in Engineering 9
Therefore, the proposedmethodology is applicable for a prac-tical application. Its contribution can be justified accordingly.
5. Conclusion
This paper proposes a procedure for solvingmultiple attributegroup decision making problems. In the proposed proce-dure, the transformation of assessment type, the propertyof consistency, the heterogeneity of a group of experts, thedetermination of weight, and scoring of alternatives are allconsidered. It would be a useful tool for decision makers indifferent industries. A review of the literature related to thisresearch suggests that no previous research has addressedall of the issues simultaneously. The proposed procedure hasseveral important properties as follows.
(i) Experts can provide their preference relations invarious formats, which can then be transformed intoa standard type.
(ii) Because all preference relation types are transformedinto fuzzy preferences, and experts only providepreference relations between alternatives ๐ and ๐ + 1, itis possible to construct preference relations matrixesthat satisfy the property of additive consistency.
(iii) Experts who are highly divergent from the groupmean will have their weights reduced.
(iv) The weights of each attribute depend on the degree ofvariation; the higher the variation of the attribute, thehigher its weight.
(v) Decisionmakers can select suitableMADMmethods,such as SAW, GRA, or TOPSIS, for the final rankingstep.
In the proposed procedure all the steps are adopted inresponse to observations made in the related literature andare understood by managers who are not experts in fuzzytheory, group decision making, MADM, or similar issues. Anumerical example was described to illustrate the proposedprocedure. It was demonstrated that the proposed procedureis simple and effective and can be easily applied to othersimilar practical problems.
The proposed procedure has some weaknesses in severalof its properties. The weight of each expert depends on thedivergence of his (or her) assessment from the opinionsof other experts. Sometimes the real expert provides themost accurate assessment but is highly divergent from themean of group. This characteristic would reduce the qualityof the group decision. Moreover, the proposed procedureassumes that an attribute is quite important if the differenceof the net degree of preference among all alternatives showsa wide variation. However, if an attribute is very importantand has a relatively high weight, any small divergence inthe assessment of the attribute can influence the rankingproduced by the group decision. These weaknesses canprovide the opportunity for future work.
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper.
Acknowledgments
This work was supported, in part, by the National ScienceCouncil of Taiwan under Grants NSC-101-2221-E-131-043 andNSC-101-2221-E-006-137-MY3.
References
[1] K. Yoon and C. L. Hwang, Multiple Attribute Decision Making:An Introduction, Sage, Thousand Oaks, Calif, USA, 1995.
[2] C. L. Hwang and K. Yoon, Multiple Attribute Decision Making:Methods and Applications, vol. 186 of Lecture Notes in Economicsand Mathematical Systems, Springer, New York, NY, USA, 1981.
[3] T. L. Saaty, The Analytical Hierarchical Process, John Wiley &Sons, New York, NY, USA, 1980.
[4] A. Charnes, W. W. Cooper, and E. Rhodes, โMeasuring theefficiency of decision making units,โ European Journal of Oper-ational Research, vol. 2, no. 6, pp. 429โ444, 1978.
[5] D. Ju-Long, โControl problems of grey systems,โ Systems &Control Letters, vol. 1, no. 5, pp. 288โ294, 1982.
[6] T. Yang and C. Kuo, โA hierarchical AHP/DEA methodologyfor the facilities layout design problem,โ European Journal ofOperational Research, vol. 147, no. 1, pp. 128โ136, 2003.
[7] T. Yang and C. C. Hung, โMultiple-attribute decision mak-ing methods for plant layout design problem,โ Robotics andComputer-Integrated Manufacturing, vol. 23, no. 1, pp. 126โ137,2007.
[8] T. Yang, Y.-C. Chang, and Y.-H. Yang, โFuzzy multiple attributedecision-makingmethod for a large 300-mm fab layout design,โInternational Journal of Production Research, vol. 50, no. 1, pp.119โ132, 2012.
[9] T. Yang, Y.-F. Wen, and F.-F. Wang, โEvaluation of robustnessof supply chain information-sharing strategies using a hybridTaguchi and multiple criteria decision-making method,โ Inter-national Journal of Production Economics, vol. 134, no. 2, pp.458โ466, 2011.
[10] J.-C. Lu, T. Yang, and C.-T. Suc, โAnalysing optimum push/pulljunction point location using multiple criteria decision-makingformultistage stochastic production system,โ International Jour-nal of Production Research, vol. 50, no. 19, pp. 5523โ5537, 2012.
[11] T. Yang and J.-C. Lu, โThe use of a multiple attribute decision-making method and value streammapping in solving the pace-maker location problem,โ International Journal of ProductionResearch, vol. 49, no. 10, pp. 2793โ2817, 2011.
[12] J. C. Lu, T. Yang, and C. Y. Wang, โA lean pull systemdesign analysed by value stream mapping and multiple criteriadecision-making method under demand uncertainty,โ Interna-tional Journal of Computer IntegratedManufacturing, vol. 24, no.3, pp. 211โ228, 2011.
[13] J.Ma, J. Lu, andG. Zhang, โDecider: a fuzzymulti-criteria groupdecision support system,โ Knowledge-Based Systems, vol. 23, no.1, pp. 23โ31, 2010.
[14] F. J. Cabrerizo, I. J. Peฬrez, and E. Herrera-Viedma, โManagingthe consensus in group decisionmaking in an unbalanced fuzzylinguistic context with incomplete information,โ Knowledge-Based Systems, vol. 23, no. 2, pp. 169โ181, 2010.
-
10 Mathematical Problems in Engineering
[15] J. Guo, โHybrid multicriteria group decision making methodfor information system project selection based on intuitionisticfuzzy theory,โMathematical Problems in Engineering, vol. 2013,Article ID 859537, 12 pages, 2013.
[16] F. Chiclana, F. Herrera, and E. Herrera-Viedma, โIntegratingthree representation models in fuzzy multipurpose decisionmaking based on fuzzy preference relations,โ Fuzzy Sets andSystems, vol. 97, no. 1, pp. 33โ48, 1998.
[17] F. Chiclana, F. Herrera, and E. Herrera-Viedma, โIntegratingmultiplicative preference relations in a multipurpose decision-making model based on fuzzy preference relations,โ Fuzzy Setsand Systems, vol. 122, no. 2, pp. 277โ291, 2001.
[18] E. Herrera-Viedma, F. Herrera, and F. Chiclana, โA consensusmodel for multiperson decision making with different pref-erence structures,โ IEEE Transactions on Systems, Man, andCybernetics Part A:Systems and Humans., vol. 32, no. 3, pp. 394โ402, 2002.
[19] Z.-P. Fan, S.-H. Xiao, and G.-F. Hu, โAn optimization methodfor integrating two kinds of preference information in groupdecision-making,โ Computers and Industrial Engineering, vol.46, no. 2, pp. 329โ335, 2004.
[20] Z.-P. Fan, J. Ma, Y.-P. Jiang, Y.-H. Sun, and L. Ma, โA goalprogramming approach to group decision making based onmultiplicative preference relations and fuzzy preference rela-tions,โ European Journal of Operational Research, vol. 174, no.1, pp. 311โ321, 2006.
[21] J. Zeng, M. An, and N. J. Smith, โApplication of a fuzzy baseddecision making methodology to construction project riskassessment,โ International Journal of Project Management, vol.25, no. 6, pp. 589โ600, 2007.
[22] E. Herrera-Viedma, F. Herrera, F. Chiclana, and M. Luque,โSome issues on consistency of fuzzy preference relations,โEuropean Journal of Operational Research, vol. 154, no. 1, pp. 98โ109, 2004.
[23] A. I. Oฬlcฬงer and A. Y. Odabasฬงi, โA new fuzzy multiple attributivegroup decision making methodology and its application topropulsion/manoeuvring system selection problem,โ EuropeanJournal of Operational Research, vol. 166, no. 1, pp. 93โ114, 2005.
[24] S. Bozoฬki, โSolution of the least squares method problem ofpairwise comparison matrices,โ Central European Journal ofOperations Research (CEJOR), vol. 16, no. 4, pp. 345โ358, 2008.
[25] Y.-M. Wang and Z.-P. Fan, โFuzzy preference relations: aggre-gation and weight determination,โ Computers & IndustrialEngineering, vol. 53, no. 1, pp. 163โ172, 2007.
[26] F. E. Boran, S. Gencฬง, M. Kurt, and D. Akay, โA multi-criteriaintuitionistic fuzzy groupdecisionmaking for supplier selectionwith TOPSISmethod,โExpert Systemswith Applications, vol. 36,no. 8, pp. 11363โ11368, 2009.
[27] F. J. Cabrerizo, S. Alonso, and E. Herrera-Viedma, โA consensusmodel for group decision making problems with unbalancedfuzzy linguistic information,โ International Journal of Informa-tion Technology and Decision Making, vol. 8, no. 1, pp. 109โ131,2009.
[28] S. J. Chuu, โGroup decision-makingmodel using fuzzymultipleattributes analysis for the evaluation of advanced manufactur-ing technology,โ Fuzzy Sets and Systems, vol. 160, no. 5, pp. 586โ602, 2009.
[29] J. Lu, Y. Zhu, X. Zeng, L. Koehl, J. Ma, and G. Zhang, โA linguis-tic multi-criteria group decision support system for fabric handevaluation,โ FuzzyOptimization andDecisionMaking, vol. 8, no.4, pp. 395โ413, 2009.
[30] Z. Zhang and X. Chu, โFuzzy group decision-making for multi-format and multi-granularity linguistic judgments in qualityfunction deployment,โ Expert Systems with Applications, vol. 36,no. 5, pp. 9150โ9158, 2009.
[31] S. Cebi and C. Kahraman, โDeveloping a group decisionsupport system based on fuzzy information axiom,โKnowledge-Based Systems, vol. 23, no. 1, pp. 3โ16, 2010.
[32] J. Kacprzyk, โGroup decision making with a fuzzy linguisticmajority,โ Fuzzy Sets and Systems, vol. 18, no. 2, pp. 105โ118, 1986.
[33] J. Kacprzyk and M. Robubnes, Non-Conventional PreferenceRelations in Decision Making, Springer, Berlin, Germany, 1988.
[34] L. Kitainik, Fuzzy Decision Procedures with Binary Relations:Towards a UnifiedTheory, vol. 13, Kluwer Academic Publishers,Dordrecht, The Netherlands, 1993.
[35] T. Tanino, โFuzzy preference orderings in group decisionmaking,โ Fuzzy Sets and Systems, vol. 12, no. 2, pp. 117โ131, 1984.
[36] S. A. Orlovsky, โDecision-making with a fuzzy preferencerelation,โ Fuzzy Sets and Systems, vol. 1, no. 3, pp. 155โ167, 1978.
[37] H.M.Hsu andC. T. Chen, โAggregation of fuzzy opinions undergroup decision-making,โ Fuzzy Sets and Systems, vol. 79, no. 3,pp. 279โ285, 1996.
[38] S. M. Chen, โAggregating fuzzy opinions in the group decision-making environment,โ Cybernetics and Systems, vol. 29, no. 4,pp. 363โ376, 1998.
[39] J. Fodor and M. Roubens, Fuzzy Preference Modelling andMulticriteria Decision Support, Kluwer Academic Publishers,Dordrecht, The Netherlands, 1994.
[40] Y. Kuo, T. Yang, and G.-W. Huang, โThe use of grey relationalanalysis in solving multiple attribute decision-making prob-lems,โ Computers and Industrial Engineering, vol. 55, no. 1, pp.80โ93, 2008.
-
Submit your manuscripts athttp://www.hindawi.com
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttp://www.hindawi.com
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Journal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
CombinatoricsHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
International Journal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
The Scientific World JournalHindawi Publishing Corporation http://www.hindawi.com Volume 2014
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com
Volume 2014 Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Stochastic AnalysisInternational Journal of