Mathematical Problems in Engineering

Volume 2015, Article ID 560690, 13 pages

http://dx.doi.org/10.1155/2015/560690

## The Interval-Valued Intuitionistic Fuzzy MULTIMOORA Method for Group Decision Making in Engineering

^{1}Department of Construction Technology and Management, Vilnius Gediminas Technical University, Saulėtekio Alėja 11, LT-10223 Vilnius, Lithuania^{2}Department of Management, Khatam Institute of Higher Education, No. 30, Hakim A’zam Alley, Mollasadra Street, North Shiraz Avenue, Tehran 19941633356, Iran^{3}Department of Management, Islamic Azad University, Kashan Branch, Kashan 8715998151, Iran

Received 12 March 2015; Revised 8 May 2015; Accepted 11 May 2015

Academic Editor: David Bigaud

Copyright © 2015 Edmundas Kazimieras Zavadskas 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.

#### Abstract

Multiple criteria decision making methods have received different extensions under the uncertain environment in recent years. The aim of the current research is to extend the application of the MULTIMOORA method (Multiobjective Optimization by Ratio Analysis plus Full Multiplicative Form) for group decision making in the uncertain environment. Taking into account the advantages of IVIFS (interval-valued intuitionistic fuzzy sets) in handling the problem of uncertainty, the development of the interval-valued intuitionistic fuzzy MULTIMOORA (IVIF-MULTIMOORA) method for group decision making is considered in the paper. Two numerical examples of real-world civil engineering problems are presented, and ranking of the alternatives based on the suggested method is described. The results are then compared to the rankings yielded by some other methods of decision making with IVIF information. The comparison has shown the conformity of the proposed IVIF-MULTIMOORA method with other approaches. The proposed algorithm is favorable because of the abilities of IVIFS to be used for imagination of uncertainty and the MULTIMOORA method to consider three different viewpoints in analyzing engineering decision alternatives.

#### 1. Introduction

Multicriteria decision making (MCDM) is a growing field of operations research both from theoretical and implementation perspectives. The importance of MCDM can be drawn from Zeleny [1]: “it has become more and more difficult to see the world around us in a uni-dimensional way and to use only a single criterion when judging what we see. We always compare, rank, and order the objects of our choice with respect to multiple criteria of choice.” The MCDM field is further divided into two classes, including multiobjective decision making (MODM) and multiattribute decision making (MADM). These classes are, respectively, associated with planning and selection classification of Simon [2] for decision making problems. The main topic of this paper covers MADM.

A MADM problem can be formally characterized as the task to evaluate, compare, and rank a set of finite alternatives, options, or choices with regard to a set of finite attributes. According to Yu [3], a MADM problem is composed of the set of substitutive alternatives (1), the set of evaluation criteria (2), the outcome (or decision) matrix with regard to the alternatives scored based on the evaluation criteria (3), and the preference structure of decision making about the criterion significances or weights (4). The information about the third and fourth parts of a decision making problem is determined exactly. However, a decision maker always deals with approximate or partial information [4]. Therefore, exactness is an unrealistic assumption. Different frameworks have been proposed to handle uncertainty in practice. Liu and Lin [5] classified these frameworks into probability and statistics (1) and grey system theory and fuzzy set theory (2). Fuzzy set theory introduced by Zadeh [6] is widely applied to decision making problems. In other words, the fuzzy set theory is a generalization of ordinal sets, where a membership degree is assigned to each element of a set. Each type of uncertainty has its own characteristics and is appropriate for special cases. While probability is concerned with the occurrence of well-defined events, fuzzy sets deal with gradual concepts and describe their boundaries [7]. In fact, the fuzzy set theory is appropriate for recognition-based uncertainty that is common in decision making.

Determining a single membership degree is a difficult task for decision makers and, therefore, Grattan-Guinness [8] believes that presentation of a linguistic expression in the form of a fuzzy set is not sufficient. In 1986, Atanassov [9] introduced the notion of intuitionistic fuzzy sets (IFSs) as an extension of ordinal fuzzy sets. In addition to a membership degree of each element, IFS assigns a degree of nonmembership to each element. Later, Atanassov and Gargov [10] extended the interval-valued intuitionistic fuzzy sets (IVIFSs), where membership and nonmembership degrees are stated as closed intervals.

When it became obvious that the type 1 fuzzy sets were not always sufficient for MADM under uncertain environment, type 2 fuzzy sets, involving interval-valued as well as intuitionistic fuzzy sets, were proposed. As a result, the MULTIMOORA method was updated by using generalized interval-valued trapezoidal fuzzy numbers [11] or intuitionistic fuzzy numbers [12]. The method has been successfully applied to making economic, technological, or management decisions. The applications of various approaches in 2006–2013, including crisp and extended methods, were summarized by T. Baležentis and A. Baležentis [13]. The most recent applications cover the extended versions of the method, in particular, the fuzzy MULTIMOORA [14, 15] or MULTIMOORA based on the interval 2-tuple linguistic variables [16]. Also, Li [17] extended MULTIMOORA with hesitant fuzzy numbers, where membership degree of elements is defined over a set of different values.

The current research aims to extend the MULTIMOORA method, using the interval-valued intuitionistic fuzzy sets. Taking into account the advantages of IVIFS, the interval-valued intuitionistic fuzzy MULTIMOORA (IVIF-MULTIMOORA) method for group decision making is presented in the paper. For this reason, three parts of MULTIMOORA were extended under IVIF information and the appropriate algorithm was proposed.

The paper is organized as follows. In Section 2, the literature review is presented and then the crisp version of MULTIMOORA is briefly overviewed in Section 3. Section 4 includes a brief description of the interval-valued intuitionistic fuzzy sets. The proposed IVIF-MULTIMOORA algorithm is presented in Section 5. The solution of two numerical examples of civil engineering problems is given in Section 6, while the results obtained are compared with the data yielded by some other methods of IVIF decision making. Finally, the discussion and the conclusions are presented in Sections 7 and 8.

#### 2. Literature Review

Taking into account the flexibility and general character of IVIFS, researchers developed decision making methods under the IVIFS environment. Some researchers developed the well-known MADM methods in the IVIF form. Some authors [18–20] proposed different frameworks of TOPSIS method under the IVIF environment. Li [21, 22] developed some mathematical programming-based method to solve the MADM problems with both ratings of the alternatives on attributes and weights of attributes expressed with the help of IVIFS. Park et al. [23] presented an IVIF version of the VIKOR method. Li [24] proposed a closeness coefficient based on the nonlinear programming method for solving IVIF MADM problems. Chen et al. [25] solved the MADM problems based on the interval-valued intuitionistic fuzzy weighted average operator and newly defined fuzzy ranking method for intuitionistic fuzzy values. Yu et al. [26] introduced some IVIF aggregation operators and applied them to solving various decision making problems. Zhang and Yu [27] presented an optimization model to determine the attribute weights in an IVIF decision making problem and, then, used these weights in an extended TOPSIS to rank the alternatives. Ye [28] introduced a cosine similarity measure and a weighted cosine similarity measure to MADM problems. Meng et al. [29] also proposed two new aggregation operators, that is, the arithmetical interval-valued intuitionistic fuzzy generalized -Shapley Choquet operator and the geometric interval-valued intuitionistic fuzzy generalized -Shapley Choquet operator, and investigated their application to solving MADM problems. Meng et al. [30] used Shapley function in extending a generalized IVIF hybrid Shapley averaging operator and proposed its application to solving MADM problems. Razavi Hajiagha et al. [31] developed Complex Proportional Assessment (COPRAS) method with IVIF data. Chai et al. [32] proposed a rule-based decision model when decision information in a group decision making problem is provided as IVIF values. Wan and Dong [33] defined the possibility degree of comparison between two IVIF numbers and introduced two ordered averaging operators based on the Karnik-Mendel algorithm to solve MADM problems with IVIF information. Chen [34] developed a method based on the traditional linear assignment method for solving decision making problems in the IVIF context. Xu and Shen [35] presented the IVIF outranking choice method to solve MADM problems, while Zavadskas et al. [36] extended the IVIF weighted aggregated sum product assessment (WASPAS) method. Geetha et al. [37] proposed a new ranking method of IVIF numbers and extended its application in decision making problems. Zhang et al. [38] proposed a new definition of IVIF entropy an entropy-based MADM method. Later, Wei, and Zhang [39] proposed applying an entropy measure for IFSs and IVIFSs to assess the experts’ weights for multicriteria fuzzy group decision making. Tong and Yu [40] introduced a novel approach for ranking the alternatives based on the IVIF cross entropy and TOPSIS in the IVIF environment. Wu and Chiclana [41, 42] introduced a risk attitudinal ranking method for IVIF numbers and applied this ranking method in a MADM problem. They imposed a risk attitude parameter over IVIF numbers ordinal score and accuracy functions [43] and used this method for solving MADM problems.

As mentioned before the current paper has focused on the extension of the MULTIMOORA (Multiobjective Optimization by Ratio Analysis plus Full Multiplicative Form) method. This method belongs to the group of complete aggregation methods, based on the reference point technique. The crisp MOORA method (Multiobjective Optimization by Ratio Analysis) was presented by Brauers and Zavadskas in 2006 [44]. This method was further supplemented by the full multiplicative form. The MULTIMOORA was suggested by Brauers and Zavadskas in 2010 [45]. The method is based on the theory of dominance [46] and enables us to summarize MOORA, consisting of the ratio system and the reference point approaches, as well as the full multiplicative form. The robustness of the method has been proved [47] and it has been successfully applied to evaluate decisions by using a single approach or its combinations with other MADM methods, such as TOPSIS [48] and WASPAS [49, 50].

Several developments of MULTIMOORA for the uncertain environment have been presented. The fuzzy MULTIMOORA was suggested by Brauers et al. [52]. To overcome some drawbacks of the fuzzy set theory, the two-tuple linguistic representation method for computing with words can be applied [53]. Accordingly, the considered method has been extended for group decision making based on two tuples (MULTIMOORA-2T-G) [54, 55]. The fuzzy MULTIMOORA, based on triangular fuzzy numbers and designed for group decision making (MULTIMOORA-FG), was introduced [56, 57]. As a further modification of the method, MOORA with grey numbers was developed [58, 59].

#### 3. MULTIMOORA

The method summarizes three approaches, that is, MOORA, consisting of the ratio system and the reference point, and the full multiplicative form.

Regardless of the approach used, initial decision criteria are transformed by applying vector normalization: where is the initial criterion value, that is, the response of the alternative to objective ; ; is the number of alternatives; ; is the number of decision criteria (objectives); is a dimensionless value of a decision criterion.

The first part of the approach is based on the ratio system [44]. To calculate the relative significance, , of each alternative with respect to all objectives , the weighted normalized criteria values should be added in the case of maximization, while, in the case of minimization, the weighted normalized criteria values should be subtracted as follows:where are maximized decision criteria; are minimized decision criteria; is the weight (or relative significance) of a criterion.

The values of variables show the preference of the alternatives according to the ratio system approach.

The second part of the approach is based on the maximal objective reference point [44]. First, the desirable ideal alternative should to be established. The virtual ideal alternative consists of the best values of the considered criteria . It is formed by selecting the best criteria values from every decision alternative based on their optimization direction, that is, the maximal values from the criteria set and the minimal values from the criteria set .

All criteria values were transformed by applying vector normalization (1). Having the dimensionless values of the criteria and relative significances (weights) of the criteria , decision alternatives are ranked based on the Min-Max metric of Tchebycheff [60]:

The second part of the approach is based on the full multiplicative form, as presented by Brauers and Zavadskas [44]. The full multiplicative form for calculating the utility of the alternatives is applied as follows: where and are calculated separately for maximized decision criteria and minimized decision criteria , respectively. and are calculated as follows:

The MULTIMOORA is based on the theory of dominance [46] and summarizes the MOORA, involving the above described ratio system and the reference point approaches, as well as the full multiplicative form [45].

Robustness of the method has been verified proving that accuracy of aggregated approaches is larger comparing to accuracy of single ones [47]. The method has been widely applied to evaluate alternatives and to select the best decisions in engineering technology or management problems by using a single approach or its combinations with other MADM methods [13, 48–50]. However, the farther the more complex decisions should be made and decision makers usually face a larger amount of approximate or partial information. Accordingly, extensions of conventional MADM methods under uncertain environment for group decision making are essential. The further developed IVIF-MULTIMOORA method could be applied in innovation themes in engineering, such as sustainable building life-cycle modelling and evaluating and selecting new business models to finance, build, and manage public buildings and infrastructures.

#### 4. Interval-Valued Intuitionistic Fuzzy Sets

Zadeh [6] generalized the characteristic function of classic sets into membership function and introduced fuzzy sets, where a membership degree was assigned to each element of a fuzzy set. Later, Atanassov [9] proposed the idea of intuitionistic fuzzy sets (IFSs), when a nonmembership degree is assigned to each element of the set alongside its membership. The interval-valued intuitionistic fuzzy sets (IVIFSs) present an extended form of IFSs.

Let be the set of all closed subintervals of the interval and let be a given set. An IVIFS in was defined as , where and with condition . The intervals and denote the degree of membership and nonmembership of the element of the set . Thus, for each , and are closed intervals whose lower and upper end points are denoted by , , , and .

The IVIFS was denoted by where , . For convenience, the IVIFS value was denoted by and referred to as an interval-valued intuitionistic fuzzy number (IVIFN).

The algebraic operations were extended over IVIFSs. Let and be any two IVIFNs; then [43]

The division and subtraction operations were defined as follows, using the extension principle [61]:

Since the final rating of the alternatives is usually determined as IVIFNs, it requires their comparison. This comparison was made, using the score and accuracy functions [43]. Let be an IVIFN. Then,was called the score function of , where , whilewas referred to as the accuracy function of , where .

For and as two IVIFNs, it follows that [43](1)if , then is smaller than , ;(2)if , then(a)if , then ,(b)if , then is smaller than , .

Another requirement of group decision making is associated with the aggregation of different judgments of decision makers into a single estimate. In this case, aggregation operators on IVIFNs can be used. Let , be a collection of IVIFNs. Then, the generalized interval intuitionistic fuzzy weighted average was defined aswhere , and is the weight vector with , and . It can be shown that GIIFWA is also an IVIFN and can be calculated as follows [62]:

If , then GIIFWA is turned into the interval intuitionistic fuzzy weighted average (IIFWA).

#### 5. IVIF-MULTIMOORA for Group Decision Making

##### 5.1. IVIF-MULTIMOORA

Assume that a group of decision makers (experts) wants to appraise a set of alternatives , based on a set of criteria . Also, the weight vector , with , and is determined that shows the relative importance of different criteria on final decision. Due to incomplete and ill-defined data, the required information in the considered problem is expressed by IVIFNs.

At the first step, each decision maker expresses his/her decision matrix , as follows:where represents the IVIF performance of the alternative over the criterion from the viewpoint of th decision maker. The aim of a decision making group is to rank the alternatives.

To solve this MAGDM problem, an IVIF version of the MULTIMOORA method, called IVIF-MULTIMOORA, was extended and described in this section. Initially, the aggregation was required to transform the set of decision matrices to an aggregated decision matrix. This aggregated decision matrix had the form of . If an equal weight was assumed for decision makers, then

The aggregation operation was performed using the GIIFWA operator and considering the weight vector for different decision makers.

Since the aggregated decision matrix was available, the IVIF-MULTIMOORA method was proposed to solve the MAGDM problem. As mentioned in Section 3, the MULTIMOORA method involves three parts: the ratio system (1), the reference point approach (2), and the full multiplicative form (3).

##### 5.2. The Part of MOORA Based on the IVIF-Ratio System

The first step in the ratio system is the normalization of the decision matrix. However, since IVIFNs are commensurable numbers, normalization is not required. The set of criteria was decomposed into two subsets of the benefit criteria (where more is better) and cost criteria (where less is better). Then, the , score of alternatives was computed by adding its weighted benefit criteria and subtracting the weighted cost criteria. If the benefit criteria were ordered as and cost criteria as , then

This score could be computed by applying the IVIFNs algebraic operators and the extension principle given in Section 4. However, a slight modification could simplify the computations. Let and . Then, for each , and for each . It was clear that , and . Similarly, , and . Then,

Multiplying both sides of (18) by positive value of , (18) was transformed as follows:

Based on the IIFWA operator definition, (18) was represented as

The multiplication of and was performed using (9). In addition, the subtraction of two IVIFNs in (21) was performed by applying (16). , is relative significance of the alternative *.* The comparison of , values based on the score and the accuracy function resulted in the ranking of the alternatives based on the ratio system.

##### 5.3. The IVIF-Reference Point Portion of the MOORA Method

According to the second part of the MOORA, the maximal objective reference point approach was used. The desirable ideal alternative in the IVIF environment was the IVIF vector with the coordinates , which was formed by selecting the data from every considered decision alternative and taking into account the optimization direction of every particular criterion. To find the reference point based on the idea of the TOPSIS method, the weighted decision matrix was constructed first, where

The scalar multiplication was performed by applying (9). Then, for benefit criteria, , the reference point was constructed as follows:and for cost criteria, ,

Then, the Min-Max metric of Tchebycheff [26] was used for ranking the alternatives:

Based on (22)–(24), (25) could be simplified. The following steps were taken to determine the rankings of the alternatives, using the reference point approach.

*Step 1. *Compute for each alternative , the score function in the weighted decision matrix:

*Step 2. *Compute the score function of the reference points:

*Step 3. *Find the maximum deviation from the reference points:

*Step 4. *Rank the alternatives in the ascending order of .

##### 5.4. The IVIF-Full Multiplicative Form

Finally, the full multiplicative form was applied as follows: where denotes the overall utility of the alternative , and

Multiplication of a set of IVIFNs was performed by sequentially applying (8). To give a compact form of this multiplication, it was supposed that , was determined as an IVIFN as indicated in (17). First, , was formed by applying (9) in the same manner as previously performed in (22). Then,

Then , was also calculated in a similar way. Then, , was determined by (10) for division. The ranking of the alternatives was specified by calculating the score functions , and ranking them in a descending order.

Applying three distinct parts of MULTIMOORA, three sets of rankings of the alternatives were determined. The obtained rankings formed a ranking pool, based on which the final ranking of the alternatives could be determined. To find the final ranking of the alternatives, the theory of dominance [46] could be used. Also, the techniques, such as the Borda count method [63] or Copeland pairwise aggregation method [64], could be applied to determine the final ranking of the alternatives.

##### 5.5. The IVIF-MULTIMOORA Algorithm

MULTIMOORA is the extension of the MOORA method and the full multiplicative form of multiple-objective. The distinct parts of the MULTIMOORA method are presented in Sections 5.1 to 5.3. The algorithmic scheme of MULTIMOORA is presented in the current section. The IVIF-MULTIMOORA algorithm includes five different stages: (1) initialization, (2) IVIF-ratio system, (3) IVIF-reference point approach, (4) IVIF-full multiplicative form, and (5) final ranking. These stages are presented in Figure 1.