Neutrosophic Sets and Systems, Vol. 19, 2018
29
University of New Mexico
TOPSIS Strategy for Multi-Attribute Decision Making with Trapezoidal Neutrosophic Numbers Pranab Biswas1, Surapati Pramanik2, and Bibhas C. Giri3 Department of Mathematics, Jadavpur University, Kolkata-700032, India. E-mail: prabiswas.jdvu@gmail.com Department of Mathematics, Nandalal B.T. College, Panpur, Narayanpur-743126, India. E-mail: sura_pati@yahoo.co.in 3 Department of Mathematics, Jadavpur University, Kolkata-700032, India. E-mail: bcgiri.jumath@gmail.com *Corresponding author: sura_pati@yahoo.co.in 1
2*
Abstract. Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) is a popular strategy for MultiAttribute Decision Making (MADM). In this paper, we extend the TOPSIS strategy of MADM problems in trapezoidal neutrosophic number environment. The attribute values are expressed in terms of single-valued trapezoidal neutrosophic numbers. The weight information of attribute is incompletely known or completely unknown. Using
the maximum deviation strategy, we develop an optimization model to obtain the weight of the attributes. Then we develop an extended TOPSIS strategy to deal with MADM with single-valued trapezoidal neutrosophic numbers. To illustrate and validate the proposed TOPSIS strategy, we provide a numerical example of MADM problem.
Keywords: Single-valued trapezoidal neutrosophic number, multi-attribute decision making, TOPSIS.
extended the TOPSIS strategy for MCDM problem in intuitionistic fuzzy environment. Zhao [25] also studied TOPMulti-attribute decision making (MADM) plays an im- SIS strategy for MADM under interval intuitionistic fuzzy portant role in decision making sciences. MADM is a pro- environment and utilized the strategy in teaching quality cess of finding the best alternative that has the highest de- evaluation. Xu [19] proposed TOPSIS strategy for hesitant gree of satisfaction over the predefined conflicting attributes. fuzzy multi-attribute decision making with incomplete The preference values of alternatives are generally assessed weight information. quantitatively and qualitatively according to the nature of attributes. When the preference values are imprecise, inde- However fuzzy sets, intuitionistic fuzzy sets, hesitant fuzzy terminate or incomplete, the decision maker feels comfort to sets have some limitations to express indeterminate and inevaluate the alternatives in MADM in terms of fuzzy sets complete information in decision making process. Recently, [1], intuitionistic fuzzy sets [2], hesitant fuzzy sets [3], neu- single valued neutrosophic set (SVNS) [26] has been suctrosophic sets [4], etc., rather than crisp sets. A large number cessfully applied in MADM or multi-attribute group deciof strategies has been developed for MADM problems such sion [27-37]. SVNS [26] and interval neutrosophic set (INS) as technique for order preference by similarity to ideal solu- [38], and other hybrid neutrosophic sets have caught attention (TOPSIS) [5], PROMETHEE [6], VIKOR [7], ELEC- tion of the researchers for developing TOPSIS strategy. TRE [7, 8], AHP [9], etc. MADM problem has been studied Biswas et al. [39] developed TOPSIS strategy for multi-atextensively in fuzzy environment [10-14], intuitionistic tribute group decision making (MAGDM) for single valued neutrosophic environment. Sahin et al. [40] proposed anfuzzy environment [15-22]. other TOPSIS strategy for supplier selection in neutrosophic TOPSIS [5] is one of the sophisticated strategy for solving environment. Chi and Liu developed TOPSIS strategy to MADM. The main idea of TOPSIS is that the best alternadeal with interval neutrosophic sets in MADM problems. tive should have the shortest distance from the positive ideal Zhang and Wu [41] proposed TOPSIS strategies for MCDM solution (PIS) and the farthest distance from the negative in single valued neutrosophic environment and interval neuideal solution (NIS), simultaneously. Since its proposition, trosophic set environment where the information about criresearchers have extended the TOPSIS strategy to deal with terion weights are incompletely known or completely undifferent environment. Chen [23] extended the TOPSIS known. Ye [42] put forward TOPSIS strategy for MAGDM strategy for solving multi-criteria decision making with single-valued neutrosophic linguistic numbers. Peng et (MCDM) problems in fuzzy environment. Boran et al. [24] al. [43] presented multi-attributive border approximation area comparison (MBAC), TOPSIS, and similarity measure 1 Introduction
Pranab Biswas, Surapati Pramanik, and Bibhas C. Giri: TOPSIS strategy for MADM with trapezoidal neutrosophic numbers
30
Neutrosophic Sets and Systems, Vol. 19, 2018
approaches for neutrosophic MADM. Pramanik et al. [44] extended TOPSIS strategy for MADM in neutrosophic soft expert set environment. Different TOPSIS strategies [45-49] have been studied in different hybrid neutrosophic set environment. Single valued trapezoidal neutrosophic number (SVTrNN ) [50, 51] is another extension of single-valued neutrosophic sets. SVTrNN presents the situation, in which each element is characterized by trapezoidal number that has truth membership degree, indeterminate membership degree, and falsity membership degree. Recently, Deli and ĹžubaĹ&#x; [52] proposed a ranking strategy of single valued neutrosophic number and utilized this strategy in MADM problems. Biswas et al. [53] also proposed value and ambiguity based ranking strategy of single valued trapezoidal neutrosophic number and applied it to MADM. However, TOPSIS strategy of MADM has not been studied earlier with trapezoidal neutrosophic numbers, although these numbers effectively deal with uncertain information in MADM model. In this study, our objective is to develop an MADM model, where the attribute values assume the form of SVTrNNs and the weight information of attribute is incompletely known or completely unknown. The existing TOPSIS strategy of MADM cannot handle with such situations. Therefore, we need to extend the TOPSIS strategy in SVTrNN environment. To develop the model, we consider the following sections: Section 2 presents a preliminaries of fuzzy sets, neutrosophic sets, single-valued neutrosophic sets, and single-valued trapezoidal neutrosophic number IFS, SVNS. Section 3 contains the extended TOPSIS strategy for MADM with SVTrNNs. Section 4 presents an illustrative example. Finally, Section 5 presents conclusion and future direction research. 2 Preliminaries In this section, we review some basic definitions of fuzzy sets, neutrosophic sets, single-valued neutrosophic sets, and single-valued trapezoidal neutrosophic number. Definition 1. [1] Let X be a universe of discourse, then a fuzzy set đ??´ is defined by A  {ďƒĄ x, ď A ( x)ďƒą | x ďƒŽ X } (1) which is characterized by a membership function ď A : X ď‚Ž [0,1] , where ď A ( x) is the degree of membership of the element x to the set A .
ďƒŹ x ď€ a)w ďƒŻ bď€a ďƒŻ ďƒŻw ď A ( x)  ďƒ ďƒŻ (d ď€ x) w ďƒŻ d ď€c ďƒŻ0 ďƒŽ
a  x  b, b  x  c, c  x  d, otherwise
where đ?‘Ž, đ?‘?, đ?‘?, đ?‘‘ are real number satisfying đ?‘Ž ≤ đ?‘? ≤ đ?‘? ≤ đ?‘‘ and đ?‘¤ is the membership degree. Definition 3.[4] Let đ?‘‹ be a universe of discourse. An neutrosophic sets đ??´ over đ?‘‹ is defined by A {x, TA (x ), I A (x ), FA (x ) | x X } (2) where TA (x ) , I A (x ) and FA (x ) are real standard or nonstandard subsets of ]ď€ 0,1 [
that is TA (x) : X ď‚Ž]ď€ 0,1 [ ,
I A (x) : X ď‚Ž]ď€ 0,1 [ FA (x) : X ď‚Ž]ď€ 0,1 [. and The membership functions satisfy the following properties: ď€ 0 ď‚Ł TA (x)  I A (x)  FA (x) ď‚Ł 3.
Definition 4. [26] Let đ?‘‹ be a universe of discourse. A single-valued neutrosophic set đ??´Ěƒ in đ?‘‹ is given by A {x, TA (x ), I A (x ), FA (x ) | x X } (3) where , and TA ( x) : X ď‚Ž [0,1] I A ( x) : X ď‚Ž [0,1] FA ( x) : X ď‚Ž [0,1] with the condition 0 ď‚Ł TA ( x)  I A ( x)  FA ( x) ď‚Ł 3 for all x ďƒŽ X . The functions TA  x  , I A  x  and FA  x  represent, respectively, the truth membership function, the indeterminacy membership function and the falsity membership function of the element x to the set A. Definition 5. [50, 51] Let a is a single valued trapezoidal neutrosophic trapezoidal number (SVNTrN). Then its truth membership function is ďƒŹ x ď€ a )ta a ď‚Ł x  b, ďƒŻ bď€a ďƒŻ b ď‚Ł x ď‚Ł c, ďƒŻ ta Ta ( x)  ďƒ ďƒŻ (d ď€ x)ta c  x ď‚Ł d, ďƒŻ d ď€c ďƒŻ 0 otherwise ďƒŽ Its indeterminacy membership function is
Definition 2. [54,55] A generalized trapezoidal fuzzy number đ??´ denoted by đ??´ = (đ?‘Ž, đ?‘?, đ?‘?, đ?‘‘; đ?‘¤) is described as a fuzzy subset of the real number â„? with membership function đ?œ‡đ??´ which is defined by
Pranab Biswas, Surapati Pramanik, and Bibhas C. Giri: TOPSIS strategy for MADM with trapezoidal neutrosophic numbers
Neutrosophic Sets and Systems, Vol. 19, 2018
31
b x ( x a ia a x b, ba b x c, ia ( x ) a x c ( d x ) i a c x d, d c 0 otherwise and its falsity membership function is b x ( x a) f a a x b, ba fa b x c, a ( x) x c (d x) f a c x d , d c 0 otherwise where 0 ≤ 𝑇𝑎̃ (𝑥) ≤ 1, 0 ≤ 𝐼𝑎̃ (𝑥) ≤ 1, 0 ≤ 𝐹𝑎̃ (𝑥) ≤ 1 and 0 ≤ 𝑇𝑎̃ (𝑥) + 𝐼𝑎̃ (𝑥) + 𝐹𝑎̃ (𝑥) ≤ 3; 𝑎, 𝑏, 𝑐, 𝑑 ∈ 𝑅. Then 𝑎̃ = ([𝑎, 𝑏, 𝑐, 𝑑]; 𝑡𝑎̃ , 𝑖𝑎̃ , 𝑓𝑎̃ , ) is called a neutrosophic trapezoidal number. Definition 5. [50,51] Let 𝑎̃1 = ([𝑎1 , 𝑏1 , 𝑐1 , 𝑑1 ]; 𝑡𝑎̃1 , 𝑖𝑎̃1, 𝑓𝑎̃1, ) and 𝑎̃2 = ([𝑎2 , 𝑏2 , 𝑐2 , 𝑑2 ]; 𝑡𝑎̃2 , 𝑖𝑎̃2, 𝑓𝑎̃2 , ) be two neutrosophic trapezoidal fuzzy numbers and 𝜆 ≥ 0, then 1. 𝑎̃1 ⊕ 𝑎̃2 = ([𝑎1 + 𝑎2 , 𝑏1 + 𝑏2 , 𝑐1 + 𝑐2 , 𝑑1 + 𝑑2 ]; 𝑡𝑎̃1 + 𝑡𝑎̃2 − 𝑡𝑎̃1 𝑡𝑎̃2 , 𝑖𝑎̃1 𝑖𝑎̃2 , 𝑓𝑎̃1 𝑓𝑎̃2 ); 2. 𝑎̃1 ⊗ 𝑎̃2 = ([𝑎1 𝑎2 , 𝑏1 𝑏2 , 𝑐1 𝑐2 , 𝑑1 𝑑2 ]; 𝑡𝑎̃1 𝑡𝑎̃2 , 𝑖𝑎̃1 + 𝑖𝑎̃2 − 𝑖𝑎̃1 𝑖𝑎̃2 , 𝑓𝑎̃1 + 𝑓𝑎̃2 − 𝑓𝑎̃1 𝑓𝑎̃2 ); 3. 𝜆 𝑎̃1 = ([𝜆 𝑎1 , 𝜆 𝑏1 , 𝜆 𝑐1 , 𝜆 𝑑1 ]; 1 − (1 − 𝜆
𝜆
𝜆
𝑡𝑎̃1 ) , (𝑖𝑎̃1 ) , (𝑓𝑎̃1 ) ; 4.
𝜆
Proof: i. The distance measure 𝑑(𝑎̃1 , 𝑎̃2 ) is obviously non-negative. If 𝑎̃1 ≈ 𝑎̃2 that is for 𝑎1 = 𝑎2 , 𝑏1 = 𝑏2 , 𝑐1 = 𝑐2 , 𝑑1 = 𝑑2 , 𝑡𝑎̃1 = 𝑡𝑎̃2 , 𝑖𝑎̃1 = 𝑖𝑎̃2 , and 𝑓𝑎̃1 = 𝑓𝑎̃2 we have 𝑑(𝑎̃1 , 𝑎̃1 ) = 0. Therefore 𝑑(𝑎̃1 , 𝑎̃2 ) ≥ 0. ii. The proof of straightforward. iii. The normalized Hamming distance between 𝑎̃1 and 𝑎̃3 is defined as follows: 𝑑(𝑎̃1 , 𝑎̃3 ) |𝑎1 (2 + 𝑡𝑎̃1 − 𝑖𝑎̃1 − 𝑓𝑎̃1 ) − 𝑎3 (2 + 𝑡𝑎̃3 − 𝑖𝑎̃3 − 𝑓𝑎̃3 )| =
|
(
𝑏1 (2 + 𝑡𝑎̃1 − 𝑖𝑎̃1 − 𝑓𝑎̃1 ) − 𝑏2 (2 + 𝑡𝑎̃2 − 𝑖𝑎̃2 − 𝑓𝑎̃2 ) +𝑏2 (2 + 𝑡𝑎̃2 − 𝑖𝑎̃2 − 𝑓𝑎̃2 ) − 𝑏3 (2 + 𝑡𝑎̃3 − 𝑖𝑎̃3 − 𝑓𝑎̃3 ) 𝑐1 (2 + 𝑡𝑎̃1 − 𝑖𝑎̃1 − 𝑓𝑎̃1 ) − 𝑐2 (2 + 𝑡𝑎̃2 − 𝑖𝑎̃2 − 𝑓𝑎̃2 )
+𝑐2 (2 + 𝑡𝑎̃2 − 𝑖𝑎̃2 − 𝑓𝑎̃2 ) − 𝑐3 (2 + 𝑡𝑎̃3 − 𝑖𝑎̃3 − 𝑓𝑎̃3 )
|
|
𝑑1 (2 + 𝑡𝑎̃1 − 𝑖𝑎̃1 − 𝑓𝑎̃1 ) − 𝑑2 (2 + 𝑡𝑎̃2 − 𝑖𝑎̃2 − 𝑓𝑎̃2 ) +| | +𝑑2 (2 + 𝑡𝑎̃2 − 𝑖𝑎̃2 − 𝑓𝑎̃2 ) − 𝑑3 (2 + 𝑡𝑎̃3 − 𝑖𝑎̃3 − 𝑓𝑎̃3 ) )
+|𝑎2 (2 + 𝑡𝑎̃2 − 𝑖𝑎̃2 − 𝑓𝑎̃2 ) − 𝑎3 (2 + 𝑡𝑎̃3 − 𝑖𝑎̃3 − 𝑓𝑎̃3 )| |𝑏1 (2 + 𝑡𝑎̃1 − 𝑖𝑎̃1 − 𝑓𝑎̃1 ) − 𝑏2 (2 + 𝑡𝑎̃2 − 𝑖𝑎̃2 − 𝑓𝑎̃2 )| =
+|𝑏2 (2 + 𝑡𝑎̃2 − 𝑖𝑎̃2 − 𝑓𝑎̃2 ) − 𝑏3 (2 + 𝑡𝑎̃3 − 𝑖𝑎̃3 − 𝑓𝑎̃3 )|
1 12
|𝑐1 (2 + 𝑡𝑎̃1 − 𝑖𝑎̃1 − 𝑓𝑎̃1 ) − 𝑐2 (2 + 𝑡𝑎̃2 − 𝑖𝑎̃2 − 𝑓𝑎̃2 )| +|+𝑐2 (2 + 𝑡𝑎̃2 − 𝑖𝑎̃2 − 𝑓𝑎̃2 ) − 𝑐3 (2 + 𝑡𝑎̃3 − 𝑖𝑎̃3 − 𝑓𝑎̃3 )| |𝑑1 (2 + 𝑡𝑎̃1 − 𝑖𝑎̃1 − 𝑓𝑎̃1 ) − 𝑑2 (2 + 𝑡𝑎̃2 − 𝑖𝑎̃2 − 𝑓𝑎̃2 )| ( +|𝑑2 (2 + 𝑡𝑎̃2 − 𝑖𝑎̃2 − 𝑓𝑎̃2 ) − 𝑑3 (2 + 𝑡𝑎̃3 − 𝑖𝑎̃3 − 𝑓𝑎̃3 )| )
Definition 6. Let 𝑎̃1 = ([𝑎1 , 𝑏1 , 𝑐1 , 𝑑1 ]; 𝑡𝑎̃1 , 𝑖𝑎̃1, 𝑓𝑎̃1, ) and 𝑎̃2 = ([𝑎2 , 𝑏2 , 𝑐2 , 𝑑2 ]; 𝑡𝑎̃2 , 𝑖𝑎̃2, 𝑓𝑎̃2, ) be two neutrosophic trapezoidal fuzzy numbers, then the normalized Hamming distance between 𝑎̃1 and 𝑎̃2 is defined as follows:
|𝑎1 (2 + 𝑡𝑎̃1 − 𝑖𝑎̃1 − 𝑓𝑎̃1 ) − 𝑎2 (2 + 𝑡𝑎̃2 − 𝑖𝑎̃2 − 𝑓𝑎̃2 )| ≤
1 12
+|𝑏1 (2 + 𝑡𝑎̃1 − 𝑖𝑎̃1 − 𝑓𝑎̃1 ) − 𝑏2 (2 + 𝑡𝑎̃2 − 𝑖𝑎̃2 − 𝑓𝑎̃2 )| +|𝑐1 (2 + 𝑡𝑎̃1 − 𝑖𝑎̃1 − 𝑓𝑎̃1 ) − 𝑐2 (2 + 𝑡𝑎̃2 − 𝑖𝑎̃2 − 𝑓𝑎̃2 )| ( +|𝑑1 (2 + 𝑡𝑎̃1 − 𝑖𝑎̃1 − 𝑓𝑎̃1 ) − 𝑑2 (2 + 𝑡𝑎̃2 − 𝑖𝑎̃2 − 𝑓𝑎̃2 )| )
𝑑(𝑎̃1 , 𝑎̃2 ) = |𝑎1 (2 + 𝑡𝑎̃1 − 𝑖𝑎̃1 − 𝑓𝑎̃1 ) − 𝑎2 (2 + 𝑡𝑎̃2 − 𝑖𝑎̃2 − 𝑓𝑎̃2 )|
|𝑎2 (2 + 𝑡𝑎̃2 − 𝑖𝑎̃2 − 𝑓𝑎̃2 ) − 𝑎3 (2 + 𝑡𝑎̃3 − 𝑖𝑎̃3 − 𝑓𝑎̃3 )|
(4)
( +|𝑑1 (2 + 𝑡𝑎̃1 − 𝑖𝑎̃1 − 𝑓𝑎̃1 ) − 𝑑2 (2 + 𝑡𝑎̃2 − 𝑖𝑎̃2 − 𝑓𝑎̃2 )| )
Property 1 The normalized Hamming distance measure 𝑑(. ) of 𝑎̃1 and 𝑎̃2 satisfies the following properties: i. 𝑑(𝑎̃1 , 𝑎̃2 ) ≥ 0, ii. 𝑑(𝑎̃1 , 𝑎̃2 ) = 𝑑(𝑎̃2 , 𝑎̃1 ), iii. 𝑑(𝑎̃1 , 𝑎̃3 ) ≤ 𝑑(𝑎̃1 , 𝑎̃2 ) + 𝑑(𝑎̃2 , 𝑎̃3 ), where 𝑎̃3 = ([𝑎3 , 𝑏3 , 𝑐3 , 𝑑3 ]; 𝑡𝑎̃3 , 𝑖𝑎̃3, 𝑓𝑎̃3, ) is a SVTrNN.
|
|𝑎1 (2 + 𝑡𝑎̃1 − 𝑖𝑎̃1 − 𝑓𝑎̃1 ) − 𝑎2 (2 + 𝑡𝑎̃2 − 𝑖𝑎̃2 − 𝑓𝑎̃2 )|
(1 − 𝑓𝑎̃1 ) )
+|𝑏1 (2 + 𝑡𝑎̃1 − 𝑖𝑎̃1 − 𝑓𝑎̃1 ) − 𝑏2 (2 + 𝑡𝑎̃2 − 𝑖𝑎̃2 − 𝑓𝑎̃2 )|
+𝑎2 (2 + 𝑡𝑎̃2 − 𝑖𝑎̃2 − 𝑓𝑎̃2 ) − 𝑎3 (2 + 𝑡𝑎̃3 − 𝑖𝑎̃3 − 𝑓𝑎̃3 )
+|
𝜆
+|𝑐1 (2 + 𝑡𝑎̃1 − 𝑖𝑎̃1 − 𝑓𝑎̃1 ) − 𝑐2 (2 + 𝑡𝑎̃2 − 𝑖𝑎̃2 − 𝑓𝑎̃2 )|
𝑎1 (2 + 𝑡𝑎̃1 − 𝑖𝑎̃1 − 𝑓𝑎̃1 ) − 𝑎2 (2 + 𝑡𝑎̃2 − 𝑖𝑎̃2 − 𝑓𝑎̃2 )
+|
1 = 12
(𝑎̃) 𝜆 = ([𝑎1𝜆 , 𝑏1𝜆 , 𝑐1𝜆 , 𝑑1𝜆 ]; (𝑡𝑎̃1 ) , 1 − (1 − 𝑖𝑎̃1 ) , 1 −
1
+|𝑐1 (2 + 𝑡𝑎̃1 − 𝑖𝑎̃1 − 𝑓𝑎̃1 ) − 𝑐3 (2 + 𝑡𝑎̃3 − 𝑖𝑎̃3 − 𝑓𝑎̃3 )| ( +|𝑑1 (2 + 𝑡𝑎̃1 − 𝑖𝑎̃1 − 𝑓𝑎̃1 ) − 𝑑2 (2 + 𝑡𝑎̃3 − 𝑖𝑎̃3 − 𝑓𝑎̃3 )| )
𝜆
12
+|𝑏1 (2 + 𝑡𝑎̃1 − 𝑖𝑎̃1 − 𝑓𝑎̃1 ) − 𝑏3 (2 + 𝑡𝑎̃3 − 𝑖𝑎̃3 − 𝑓𝑎̃3 )|
1 12
+
1 12
+|𝑏2 (2 + 𝑡𝑎̃2 − 𝑖𝑎̃2 − 𝑓𝑎̃2 ) − 𝑏3 (2 + 𝑡𝑎̃3 − 𝑖𝑎̃3 − 𝑓𝑎̃3 )| +|𝑐2 (2 + 𝑡𝑎̃2 − 𝑖𝑎̃2 − 𝑓𝑎̃2 ) − 𝑐3 (2 + 𝑡𝑎̃3 − 𝑖𝑎̃3 − 𝑓𝑎̃3 )| ( +|𝑑2 (2 + 𝑡𝑎̃2 − 𝑖𝑎̃2 − 𝑓𝑎̃2 ) − 𝑑3 (2 + 𝑡𝑎̃3 − 𝑖𝑎̃3 − 𝑓𝑎̃3 )| )
≤ 𝑑(𝑎̃ 1 , 𝑎̃ 2 ) + 𝑑(𝑎̃ 2 , 𝑎̃ 3 ) .
□
2.1 TOPSIS Strategy for MADM The idea behind the TOPSIS strategy [5] is to find out the optimal alternative that has the shortest distance from the positive ideal solution and the farthest distance from the
Pranab Biswas, Surapati Pramanik, and Bibhas C. Giri: TOPSIS strategy for MADM with trapezoidal neutrosophic numbers
32
Neutrosophic Sets and Systems, Vol. 19, 2018
negative ideal solution, simultaneously. The schematic structure of classical TOPSIS strategy is presented in the following figure (see Fig. 1)
Construct a decision matrix
Normalize the decision matrix
Calculate the weighted normalized decision matrix
Determine the positive and negative ideal solutions
Calculate the distance measure of each alternative from ideal solution
Calculate relative closeness co-efficients of the alternatives
Rank the alternatives Figure 1. A schematic structure of TOPSIS strategy 3 TOPSIS strategy for multi-attribute decision making with neutrosophic trapezoidal number In this section, we put forward a framework for determining the attribute weights and the ranking orders for all the alternatives with incomplete weight information under neutrosophic environment. Consider a MADM problem, where đ??´ = {đ??´1 , đ??´2 , ‌ , đ??´đ?‘š } is a set of đ?‘š alternatives and đ??ś = {đ??ś1 , đ??ś2 , ‌ , đ??śđ?‘› } is a set of đ?‘› attributes. The attribute value of alternative đ??´đ?‘– (đ?‘– = 1,2, ‌ , đ?‘š) over the attribute đ??śđ?‘— (đ?‘— = 1,2, ‌ , đ?‘›) assumes the form of neutrosophic trapezoidal number đ?‘ŽĚƒđ?‘–đ?‘— = ([đ?‘Žđ?‘–đ?‘— , đ?‘?đ?‘–đ?‘— , đ?‘?đ?‘–đ?‘— , đ?‘‘đ?‘–đ?‘— ]; đ?‘Ąđ?‘ŽĚƒđ?‘–đ?‘— , đ?‘–đ?‘ŽĚƒđ?‘–đ?‘—, đ?‘“đ?‘ŽĚƒđ?‘–đ?‘—, ), where 0 ≤ đ?‘Ąđ?‘ŽĚƒđ?‘–đ?‘— ≤ 1, 0 ≤
đ?‘–đ?‘ŽĚƒđ?‘–đ?‘— ≤ 1 , 0 ≤ đ?‘“đ?‘ŽĚƒđ?‘–đ?‘— ≤ 1 and 0 ≤ đ?‘Ąđ?‘ŽĚƒđ?‘–đ?‘— + đ?‘–đ?‘ŽĚƒđ?‘–đ?‘— + đ?‘“đ?‘ŽĚƒđ?‘–đ?‘— ≤ 3; đ?‘Ž, đ?‘?, đ?‘?, đ?‘‘ ∈ đ?‘…. Here, đ?‘Ąđ?‘ŽĚƒđ?‘–đ?‘— denotes the truth membership degree, đ?‘–đ?‘ŽĚƒđ?‘–đ?‘— denotes the indeterminate membership degree, and đ?‘“đ?‘ŽĚƒđ?‘–đ?‘— denotes the falsity membership degree to consider the trapezoidal number [đ?‘Žđ?‘–đ?‘— , đ?‘?đ?‘–đ?‘— , đ?‘?đ?‘–đ?‘— , đ?‘‘đ?‘–đ?‘— ] as the rating values of đ??´đ?‘– over the attribute đ??śđ?‘— . An MADM problem can be expressed by a decision matrix in which the entries represent the evaluation information of all alternatives with respect to the attributes. Then we construct the following neutrosophic decision matrix, whose elements are SVNTrNs: đ?‘ŽĚƒ11 đ?‘ŽĚƒ12 ‌ đ?‘ŽĚƒ1đ?‘› đ?‘ŽĚƒ21 đ?‘ŽĚƒ22 ‌ đ?‘ŽĚƒ2đ?‘› đ??ˇ = (đ?‘ŽĚƒđ?‘–đ?‘— )đ?‘šĂ—đ?‘› = ( ) (5) â‹Ž â‹Ž â‹ą â‹Ž đ?‘ŽĚƒđ?‘š1 đ?‘ŽĚƒđ?‘š2 ‌ đ?‘ŽĚƒđ?‘šđ?‘› Due to different attribute weights, we assume that the weight vector of all attributes is given by đ?‘¤ = (đ?‘¤1 , đ?‘¤2 , ‌ , đ?‘¤đ?‘› )đ?‘‡ , where 0 ≤ đ?‘¤đ?‘— ≤ 1, đ?‘— = 1,2, ‌ , đ?‘› , and đ?‘¤đ?‘— is the weight of each attribute. The information about attribute weights is usually incomplete in decision making problems under uncertain environment. For convenience, we assume ď „ be a set of the known weight information [56-59], where ď „ can be constructed by the following forms, for đ?‘– ≠đ?‘—: Form 1. A weak ranking: {đ?‘¤đ?‘– ≼ đ?‘¤đ?‘— }; Form 2. A strict ranking: {đ?‘¤đ?‘– − đ?‘¤đ?‘— ≼ đ?›źđ?‘— }(đ?›źđ?‘— > 0); Form 3. A ranking of difference: {đ?‘¤đ?‘– − đ?‘¤đ?‘— ≼ đ?‘¤đ?‘˜ − đ?‘¤đ?‘™ }, for đ?‘— ≠đ?‘˜ ≠đ?‘™; Form 4. A ranking with multiples: {đ?‘¤đ?‘– ≼ đ?›źđ?‘— đ?‘¤đ?‘— } (0 ≤ đ?›źđ?‘— ≤ 1); Form 5. An interval form: {đ?›źđ?‘– ≤ đ?‘¤đ?‘– ≤ đ?›źđ?‘– + đ?œ–đ?‘– }(0 ≤ đ?›źđ?‘— ≤ đ?›źđ?‘– + đ?œ–đ?‘– ≤ 1). Now we develop a strategy for solving the MADM problems, in which the information about attribute weights is completely unknown or partially known and the attribute values are expressed by SVTrNNs. The following steps are considered to develop the model. 3.1 Standardize the decision matrix Let D   aij mď‚´n be a neutrosophic decision matrix, where


the SVTrNNs aij  ďƒŠďƒŤaij1 , aij2 , aij3 , aij4 ďƒšďƒť ; ta , ia , f a is the rating values of alternative Ai with respect to attribute C j . Now to ij
ij
ij
eliminate the effect from different physical dimensions into decision making process, we should standardize the decision matrix  aij m n based on two common types of attributes such as benefit type attribute and cost type attribute. We consider the following technique to obtain the
Pranab Biswas, Surapati Pramanik, and Bibhas C. Giri: TOPSIS strategy for MADM with trapezoidal neutrosophic numbers
33
Neutrosophic Sets and Systems, Vol. 19, 2018
standardized decision matrix R   rij m n , in which the

component rijk of the entry rij  ďƒŠďƒŤ rij1 , rij2 , rij3 , rij4 ďƒšďƒť ; tr , ir , f r the matrix R are considered as: 1. For benefit type attribute: ij
ij
ij

đ?‘š
2 ij  j
3 ij  j
4 ij  j
(6)
(7)
r1n ďƒś ďƒˇ r2 n ďƒˇ ďƒˇ ďƒˇ rmn ďƒˇďƒ¸
(8)
To determine the attribute weights, we use maximum deviation strategy, which was proposed by Wang [60]. According to Wang [60], i. The attribute that has the larger deviation value among alternatives should be assigned larger weight. ii. The attribute having deviation value among alternatives should be assigned smaller weight. iii. The attribute having no deviation among alternatives should be assigned zero weight. Following the idea of maximum deviation method, we construct an optimization model to determine the optimal weights of attributes with SVTrNNs. The deviation of the alternative đ??´đ?‘– to all the other alternatives for the attribute C j can be defined as follows: đ?‘– = 1,2, ‌ , đ?‘š;
đ?‘—=
đ?‘‘(đ?‘ŽĚƒđ?‘–đ?‘— , đ?‘ŽĚƒđ?‘˜đ?‘— ) = |đ?‘Žđ?‘–đ?‘—1 (2 + đ?‘Ąđ?‘ŽĚƒđ?‘–đ?‘— − đ?‘–đ?‘ŽĚƒđ?‘–đ?‘— − đ?‘“đ?‘ŽĚƒđ?‘–đ?‘— ) − đ?‘Žđ?‘˜đ?‘—1 (2 + đ?‘Ąđ?‘ŽĚƒđ?‘˜đ?‘— − đ?‘–đ?‘ŽĚƒđ?‘˜đ?‘— − đ?‘“đ?‘ŽĚƒđ?‘˜đ?‘— )| + |đ?‘Žđ?‘–đ?‘—2 (2 + đ?‘Ąđ?‘ŽĚƒđ?‘–đ?‘— − đ?‘–đ?‘ŽĚƒđ?‘–đ?‘— − đ?‘“đ?‘ŽĚƒđ?‘–đ?‘— ) − đ?‘Žđ?‘˜đ?‘—2 (2 + đ?‘Ąđ?‘ŽĚƒđ?‘˜đ?‘— − đ?‘–đ?‘ŽĚƒđ?‘˜đ?‘— − đ?‘“đ?‘ŽĚƒđ?‘˜đ?‘— )|
1 12
+ |đ?‘Žđ?‘–đ?‘—3 (2 + đ?‘Ąđ?‘ŽĚƒđ?‘–đ?‘— − đ?‘–đ?‘ŽĚƒđ?‘–đ?‘— − đ?‘“đ?‘ŽĚƒđ?‘–đ?‘— ) − đ?‘Žđ?‘˜đ?‘—3 (2 + đ?‘Ąđ?‘ŽĚƒđ?‘˜đ?‘— − đ?‘–đ?‘ŽĚƒđ?‘˜đ?‘— − đ?‘“đ?‘ŽĚƒđ?‘˜đ?‘— )|
(9)
( + |đ?‘Žđ?‘–đ?‘—4 (2 + đ?‘Ąđ?‘ŽĚƒđ?‘–đ?‘— − đ?‘–đ?‘ŽĚƒđ?‘–đ?‘— − đ?‘“đ?‘ŽĚƒđ?‘–đ?‘— ) − đ?‘Žđ?‘˜đ?‘—4 (2 + đ?‘Ąđ?‘ŽĚƒđ?‘˜đ?‘— − đ?‘–đ?‘ŽĚƒđ?‘˜đ?‘— − đ?‘“đ?‘ŽĚƒđ?‘˜đ?‘— )| )
=
1 12
∑4đ?‘?=1 |
đ?‘Žđ?‘–đ?‘—đ?‘? (2 + đ?‘Ąđ?‘ŽĚƒđ?‘–đ?‘— − đ?‘–đ?‘ŽĚƒđ?‘–đ?‘— − đ?‘“đ?‘ŽĚƒđ?‘–đ?‘— ) −đ?‘Žđ?‘˜đ?‘—đ?‘? (2 + đ?‘Ąđ?‘ŽĚƒđ?‘˜đ?‘— − đ?‘–đ?‘ŽĚƒđ?‘˜đ?‘— − đ?‘“đ?‘ŽĚƒđ?‘˜đ?‘— )
đ?‘›
đ?‘›
đ?‘š
đ??ˇ(đ?‘¤) = ∑ đ??ˇđ?‘— (đ?‘¤) = ∑ ∑ đ?‘‘đ?‘–đ?‘— (đ?‘¤) đ?‘—=1 đ?‘–=1
đ?‘š 4 =∑đ?‘›đ?‘—=1 ∑đ?‘š đ?‘–=1 ∑đ?‘˜=1 (12 ∑đ?‘?=1 | 1
đ?‘Žđ?‘?đ?‘–đ?‘— (2 + đ?‘Ąđ?‘ŽĚƒđ?‘–đ?‘— − đ?‘–đ?‘ŽĚƒđ?‘–đ?‘— − đ?‘“đ?‘ŽĚƒđ?‘–đ?‘— ) đ?‘?
−đ?‘Žđ?‘˜đ?‘— (2 + đ?‘Ąđ?‘ŽĚƒđ?‘˜đ?‘— − đ?‘–đ?‘ŽĚƒđ?‘˜đ?‘— − đ?‘“đ?‘ŽĚƒđ?‘˜đ?‘— )
|) ��
If the information about the attribute weights is partially known or completely unknown, then we propose two models to obtain the attribute weights. 3.2.1 Information about the weights of attributes is partially known.
3.2 Determine the attribute weight
,
12
đ?‘—=1
Then we obtain the following standardized decision matrix:
đ?‘‘đ?‘–đ?‘— (đ?‘¤) = ∑đ?‘š Ěƒđ?‘–đ?‘— , đ?‘ŽĚƒđ?‘˜đ?‘— )đ?‘¤đ?‘— đ?‘˜=1 đ?‘‘( đ?‘Ž 1,2, ‌ , đ?‘› where
đ?‘–=1 đ?‘˜=1
1
đ?‘š đ?‘š = ∑đ?‘›đ?‘—=1 ∑đ?‘š Ěƒđ?‘–đ?‘— , đ?‘ŽĚƒđ?‘˜đ?‘— ) đ?‘¤đ?‘— đ?‘–=1 ∑đ?‘–=1 ∑đ?‘˜=1 đ?‘‘(đ?‘Ž
u ď€j  min{aij1 | i  1, 2,...m} for j  1,2,...n.
mď‚´ n
đ?‘–=1
đ?‘? đ?‘š 4 = ∑đ?‘š đ?‘–=1 ∑đ?‘˜=1 ( ∑đ?‘?=1 |đ?‘Žđ?‘–đ?‘— (2 + đ?‘Ąđ?‘ŽĚƒđ?‘–đ?‘— − đ?‘–đ?‘ŽĚƒđ?‘–đ?‘— − đ?‘?
 4 where u j  max{aij | i  1, 2,...m} and
R   rij 
đ?‘š
đ?‘“đ?‘ŽĚƒđ?‘–đ?‘— ) − đ?‘Žđ?‘˜đ?‘— (2 + đ?‘Ąđ?‘ŽĚƒđ?‘˜đ?‘— − đ?‘–đ?‘ŽĚƒđ?‘˜đ?‘— − đ?‘“đ?‘ŽĚƒđ?‘˜đ?‘— )|) đ?‘¤đ?‘— . (10) Similarly, the deviation value of all the alternatives to other alternatives for all the criteria can be taken as:
ďƒŹďƒŻďƒŚ ďƒŠ u ď€j u ď€j u ď€j u ď€j ďƒš ďƒś rij  ďƒďƒ§ ďƒŞ 4 , 3 , 2 , 1 ďƒş ; trij , irij , f rij ďƒˇ , ďƒ§ ďƒˇ a a a a ďƒ¸ ďƒŽďƒŻďƒ¨ ďƒŞďƒŤ ij ij ij ij ďƒşďƒť
ďƒŚ r11 r12 ďƒ§ r r  ďƒ§ 21 22 ďƒ§ ďƒ§ďƒ§ ďƒ¨ rm1 rm 2
đ?‘š
đ??ˇđ?‘— (đ?‘¤) = ∑ đ?‘‘đ?‘–đ?‘— (đ?‘¤) = ∑ ∑ đ?‘‘(đ?‘ŽĚƒđ?‘–đ?‘— , đ?‘ŽĚƒđ?‘˜đ?‘— ) đ?‘¤đ?‘—
ďƒŹ ďƒś ďƒŻďƒŚ ďƒŠ a a a a ďƒš rij  ďƒďƒ§ ďƒŞ , , , ďƒş ; trij , irij , f rij ďƒˇ ďƒ§ u u u u ďƒş ďƒˇ ďƒŻ ďƒť ďƒ¸ ďƒŽďƒ¨ ďƒŤďƒŞ 2. For cost type attribute: 1 ij  j
in
denotes the neutrosophic Hamming distance between two SVTrNNs đ?‘ŽĚƒđ?‘–đ?‘— and đ?‘ŽĚƒđ?‘˜đ?‘— . The deviation value of all the alternatives to other alternatives for the attribute đ??śđ?‘— can be obtained as follows:
|
In order to obtain the weight vector, we construct a non-linear programming model that maximizes all deviation values of attributes. The model can be presented as follows: p ďƒŹ ďƒś n m m 4 ďƒŚ a (2  t a ď€ ia ď€ f a ) ij ij ij ďƒŻmax D  w  1 ďƒĽďƒĽďƒĽďƒĽ ďƒ§ ij ďƒˇ wj ďƒŻďƒŻ 12 j 1 i 1 k 1 p 1 ďƒ§ ď€akjp (2  ta ď€ ia ď€ f a ) ďƒˇ kj kj kj ďƒ¨ ďƒ¸ ( M ď€ 2) ďƒ n ďƒŻ ďƒŻsubject to w ďƒŽ ď „, ďƒĽ w j  1, w j ď‚ł 0, for j  1,2,.., n. ďƒŻďƒŽ j 1
(11) By solving the model (M-1), we obtain the optimal solution to be used as the weight vector. 3.2.2 Information about the weights of attributes is unknown.
If the information about attribute weight is completely unknown, then we can establish the following programming model: p ďƒŹ ďƒś n m m 4 ďƒŚ a (2  t a ď€ ia ď€ f a ) ij ij ij ďƒŻmax D  w  1 ďƒĽďƒĽďƒĽďƒĽ ďƒ§ ij ďƒˇ wj p 12 j 1 i 1 k 1 p 1 ďƒ§ ď€akj (2  ta ď€ ia ď€ f a ) ďƒˇ ďƒŻďƒŻ kj kj kj ďƒ¨ ďƒ¸ ( M ď€ 2) ďƒ n ďƒŻ 2 ďƒŻsubject to w ďƒŽ ď „, ďƒĽ w j  1, w j ď‚ł 0, for j  1,2,.., n. ďƒŻďƒŽ j 1
(12) To solve the model (M-2), we develop the Lagrange function:
Pranab Biswas, Surapati Pramanik, and Bibhas C. Giri: TOPSIS strategy for MADM with trapezoidal neutrosophic numbers
34
Neutrosophic Sets and Systems, Vol. 19, 2018
L w,
w w1 , w2 ,..., wn .
p 1 n m m 4 aij (2 taij iaij f aij ) wj 12 j 1 i 1 k 1 p 1 akjp (2 ta ia f a ) kj kj kj n 2 w j 1 24 j 1
(13)
where is a real number and denoting the Lagrange multiplier variable. Then the partial derivative of L with respect to wj ( j 1,2,..., n) and are obtained as: p m m 4 aij (2 ta ia f a ) L ij ij ij wj wj 0 p w j i 1 k 1 p 1 akj (2 ta ia f a ) kj kj kj n L wj2 1 0 j 1
(14) (15)
It follows from Eq. (14) that m
for
j 1,2,..., n.
(16)
Putting the values of w j in Eq.(15), we obtain m m 4 aijp (2 ta ia f a ) ij ij ij p j 1 i 1 k 1 p 1 akj (2 t a ia f a ) kj kj kj
2
n
(17)
m m 4 aijp (2 ta ia f a ) ij ij ij p j 1 i 1 k 1 p 1 akj (2 t a ia f a ) kj kj kj
2
n
for 0.
(18) Then combining Eq.(16) and Eq.(18), we obtain the following formula for determining the weight of attribute C j ( j 1,2,..., n) : m
wj
m
4
aijp (2 ta ia f a ) ij ij ij
a i 1 k 1 p 1
p kj (2 t akj iakj f akj )
m m 4 aijp (2 ta ia f a ) ij ij ij p j 1 i 1 k 1 p 1 akj (2 t a ia f a ) kj kj kj
2
.
(19)
rj rj1 , rj 2 , rj 3 , rj 4 , t j , i j , f j
max r , max r , max r , max r 4 ; ij i i i i max tij , min iij , min f ij i i i 1 ij
2 ij
3 ij
(22)
min r1 , min r 2 , min r 3 , min r 4 ; ij ij ij ij i i i i min tij , max iij , max f ij i i i
(23)
Moreover, the trapezoidal neutrosophic global positive ideal solution and the trapezoidal neutrosophic global trapezoidal global negative ideal solution can be directly considered as rj 1,1,1,1 ,1,0,0 and rj 0,0,0,0,0,1,1 (24) 3.4 Determine the separation measures from ideal solutions to each alternative The separation measures di and di of each alternative from the ideal solutions can be determined by Eq.(9), Eq.(20) and Eq.(21), respectively, as follows: di w j d rij , rj n
n
C j ( j 1,2,..., n) : aijp (2 ta ia f a ) ij ij ij p a (2 t i f akj ) i 1 k 1 p 1 kj akj akj wj wj n p n m m 4 a (2 t i f ) j 1 w j aij aij aij ij p j 1 i 1 k 1 p 1 akj (2 t a ia f a ) kj kj kj
p 4 rij (2 t r ir f r ) ij ij ij 1 n for i 1,2,..., m (25) wj 12 j 1 p 1 rjp (2 tr ir f r ) j j j
di w j d rij , rj
We make their sum into a unit by normalizing w j ( j 1,2,..., n) and get the optimal weight of attribute
m
where,
j 1
n
m
neutrosophic trapezoidal local positive ideal solution (NTrPIS) and the neutrosophic trapezoidal local negative ideal solution (NTrNIS) are defined as follows r r1 , r2 ,..., rn and r r1 , r2 ,..., rn (21)
4
2
In the normalized decision matrix R rij m n , the
rj rj1 , rj 2 , rj 3 , rj 4 , t j , i j , f j
aijp (2 ta ia f a ) ij ij ij p i 1 k 1 p 1 akj (2 ta ia f a ) kj kj kj wj m
3.3 Determine the ideal solutions
4
j 1
p 4 rij (2 t r ir f r ) ij ij ij 1 n for i 1,2,..., m (26) wj p 12 j 1 p 1 rj (2 tr ir f r ) j j j
3.5 Determine the relative closeness co-efficient (20)
The relative closeness co-efficient of an alternative Ai with respect to ideal alternative A is defined as the following formula:
Then we get the normalized weight vector of attributes:
Pranab Biswas, Surapati Pramanik, and Bibhas C. Giri: TOPSIS strategy for MADM with trapezoidal neutrosophic numbers
35
Neutrosophic Sets and Systems, Vol. 19, 2018
RC ( Ai )
d i d i d i
(27)
where 0 RC( Ai ) 1 for i 1,2,...m. According to the closeness co-efficient RC( Ai ) , the ranking orders of all alternatives and the best alternative can be selected. The schematic diagram of the proposed TOPSIS is presented in Figure-2.
2.
Past experience (C 2 ) ,
3.
General aptitude (C 3 ) and
4.
Self- confidence (C 4 ) .
The decision maker evaluates the ratings of alternatives Ai (i 1,2,..., m) with respect to the attributes Ci (i 1,2,..., n) with the decision matrix D (aij )44 (see Table 1). Table 1. Rating values of alternatives
Construct a decision matix
Problem formulation
Standardize the decision matrix
Determine the attribute weights
The maximum deviation strategy
Determine the ideal solutions
Determine the separation measures
Calculate the relative closeness co-efficients
TOPSIS strategy with SVTrNNs
Select the best alternative
Figure 2. The schematic diagram of the proposed startegy 4 An illustrative example In this section, we consider an illustrative example of medical representative selection problem to demonstrate and applicability of the proposed. Consider a MADM problem, where a pharmacy company wants to recruit a medical representative. After initial scrutiny four candidates Ai (i 1,2, 3, 4) have been consid-
C1
C2
A1
[7,8,9,10]; 0.90, 0.10, 0.05
5, 6, 7,8 ; 0.65, 0.35, 0.30
A2
5, 6, 7,8 ; 0.65, 0.35, 0.30
6, 7,8,9 ; 0.80, 0.20, 0.15
A3
4,5, 6, 7 ; 0.50, 0.50, 0.45
5, 6, 7,8 ; 0.65, 0.35, 0.30
A4
4,5, 6, 7 ; 0.50, 0.50, 0.45
5, 6, 7,8 ; 0.65, 0.35, 0.30
C3
C4
A1
6, 7,8,9 ; 0.80, 0.20, 0.15
7,8,9,10 ; 0.90, 0.10, 0.05
A2
7,8,9,10 ; 0.90, 0.10, 0.05
6, 7,8,9 ; 0.80, 0.20, 0.15
A3
4,5, 6, 7 ; 0.50, 0.50, 0.45
4,5, 6, 7 ; 0.50, 0.50, 0.45
A4
4,5, 6, 7 ; 0.50, 0.50, 0.45
6, 7,8,9 ; 0.80, 0.20, 0.15
The information of the attributes is incompletely known and the weight information is given as follows: 0.20 w1 0.30,0.05 w2 0.20, 4 0.20 w 0.35,0.15 w 0.35; w 1 3 4 j j 1
To determine the best alternative, we use the proposed strategy involving the following steps: Step 1. Standardize the decision matrix Since the selective attributes are benefit type attributes, then using Eq. (6), we have the following standardized decision matrix: R (rij )44 (see Table 2.) Table 2. Standardized rating values of alternatives C1
C2 0.5, 0.6, 0.7, 0.8 ; 0.65, 0.35, 0.30 0.6, 0.7, 0.8, 0.9; 0.80, 0.20, 0.15
ered for further evaluation with respect to the four attributes C j (j 1,2, 3, 4) namely,
A1
[0.7, 0.8, 0.9,1.0]; 0.90, 0.10, 0.05
1.
A2
0.5, 0.6, 0.7, 0.8 ; 0.65, 0.35, 0.30
Oral communication skill (C 1 ) ;
(28)
Pranab Biswas, Surapati Pramanik, and Bibhas C. Giri: TOPSIS strategy for MADM with trapezoidal neutrosophic numbers
36
Neutrosophic Sets and Systems, Vol. 19, 2018
A3
0.4, 0.5, 0.6, 0.7 ; 0.50, 0.50, 0.45
0.5, 0.6, 0.7, 0.8 ; 0.65, 0.35, 0.30
Step 4. Determine the separation measures from ideal solutions to each alternative.
A4
0.4, 0.5, 0.6, 0.7 ; 0.50, 0.50, 0.45
0.5, 0.6, 0.7, 0.8 ; 0.65, 0.35, 0.30
Case 1. Employing Eq.(25), we obtain the separation measures di of each alternative Ai (i 1,2,3,4) from A :
C3
C4
A1
0.6, 0.7, 0.8, 0.9; 0.80, 0.20, 0.15
0.7, 0.8, 0.9,1.0 ; 0.90, 0.10, 0.05
A2
0.7, 0.8, 0.9,1.0 ; 0.90, 0.10, 0.05
0.6, 0.7, 0.8, 0.9; 0.80, 0.20, 0.15
A3
0.4, 0.5, 0.6, 0.7 ; 0.50, 0.50, 0.45
0.4, 0.5, 0.6, 0.7 ; 0.50, 0.50, 0.45
A4
0.4, 0.5, 0.6, 0.7 ; 0.50, 0.50, 0.45
0.6, 0.7, 0.8, 0.9; 0.80, 0.20, 0.15
Similarly, using Eq.(26), we obtain the separation measures di of each alternative Ai (i 1,2,3,4) from A : d A1 , A 0.4119 ,
d A2 , A 0.3254 ,
d A3 , A 0 ,
d A4 , A 0.0985.
Case 2. The separation measures di of each alternative Ai (i 1,2,3,4) from A : d A4 , A 0.3708.
Case 1. Weight information is incompletely known. Using the model (M-1), we construct the following singleobjective programming problem: max D w 3.2133w1 1.1401w2 3.4250w3 2.9700w4 4 (29) subject to w , w j 1, w j 0, for j 1,2,..,4. j 1
Solving this model with optimization software LINGO 13, we get the optimal weight vector as w 0.30,0.05,0.35,0.30 .
Case 2. Weight information is completely unknown. Following Eq.(20), we obtain the following optimal weight vector: w 0.2990,0.1061,0.3186,0.2763 .
Step 3. Determine the ideal solutions Since the chosen attributes are benefit type attribute, then following Eq.(22) we determine the neutrosophic trapezoidal positive ideal solution as (30)
Similarly, using Eq.(23), we determine the neutrosophic trapezoidal negative ideal solution 0.4,0.5,0.6,0.7;0.50,0.50,0.45 , 0.5,0.6,0.7,0.8;0.65,0.35,0.30 , A 0.4,0.5,0.6,0.7;0.50,0.50,0.45 , 0.4,0.5,0.6,0.7;0.50,0.50,0.45
d A4 , A 0.3807.
d A1 , A 0.0721 , d A2 , A 0.1494 , d A3 , A 0.4615 ,
Step 2. Determine the attribute weight
[0.7,0.8,0.9,1.0];0.90,0.10,0.05 , [0.6,0.7,0.8,0.9];0.80,0.20,0.15 , A [0.7,0.8,0.9,1.0];0.90,0.10,0.05 , [0.7,0.8,0.9,1.0];0.90,0.10,0.05
d A1 , A 0.0673 , d A2 , A 0.1538 , d A3 , A 0.4792 ,
(31)
Similarly, the separation measures di of each alternative Ai (i 1,2,3,4) from A : d A1 , A 0.3894 ,
d A2 , A 0.3120 ,
d A3 , A 0 ,
d A4 , A 0.0907.
Step 5. Calculate the relative closeness coefficient. Using Eq.(27), we calculate the relative closeness coefficient RC( Ai ) of alternative Ai (i 1,2,3,4) for Case 1 and Case 2, respectively. We put the result in Table 3. Table 3. Rating values of alternatives RC(Ai) Case 1 Case 2 RC(A1) 0.8596 0.8438 RC(A2) 0.6790 0.6824 RC(A3) 0 0 RC(A4) 0.2056 0.1965 Following Table 3, we rank the alternatives Ai (i 1,2,3,4) according to the values of relative closeness coefficient RC( Ai ) for both cases: A1 A2 A4 A3. Therefore A1 is the best alternative.
5 Conclusions TOPSIS strategy is a useful strategy for solving MADM problem under different environment. In this paper, we have investigated MADM problems with SVTrNNs. The weight
Pranab Biswas, Surapati Pramanik, and Bibhas C. Giri: TOPSIS strategy for MADM with trapezoidal neutrosophic numbers
37
Neutrosophic Sets and Systems, Vol. 19, 2018
information of attributes have been considered to be incompletely known or completely unknown. First, we have used Hamming distance measure to determine the distance measure of SVTrNNs. Second, we developed an optimization model to determine the attribute weights based on the idea of maximum deviation strategy. Third, we have extended the TOPSIS strategy for solving the MADM model with SVTrNNs. Finally, we have provided an illustrative examples to verify the feasibility and effectiveness of the proposed model. The proposed TOPSIS strategy can be extended to multi-attribute group decision making with SVTrNNs and multi-attribute decision making problem with interval trapezoidal neutrosophic numbers. The proposed TOPSIS strategy can be used in solving logistics center location selection [61, 62], weaver selection [63, 64], data mining [65], school choice [66], teacher selection [67], brick field selection [68-69), etc. under SVTrNN environment. References [1]. L.A. Zadeh. Fuzzy sets. Information and Control, 8 (3) (1965), 338–353. [2]. K. Atanassov. More on intuitionistic fuzzy sets. Fuzzy Sets and Systems, 33 (1989), 37–46. [3]. V. Torra, Hesitant fuzzy sets. International Journal of Intelligent Systems, 25 (2010), 529 – 539. [4]. F. Smarandache, A unifying field in logics. Neutrosophy: neutrosophic probability, set and logic. American Research Press, Rehoboth 1998. [5]. C.L. Hwang, and K. Yoon. Multiple attribute decision making, methods and applications. Springer-Verlag, New York 1981. [6]. J.P. Brans, P. Vincke, and B. Mareschal. How to select and how to rank projects: The PROMETHEE method. European Journal of Operation Research, 24 (1986), 228–238. [7]. S. Opricovic, and G. H. Tzeng. Compromise solution by MCDM methods: A comparative analysis of VIKOR and TOPSIS. European Journal of Operation Research, 156 (2004), 445–455. [8]. B. Roy. The outranking approach and the foundations of ELECTRE methods. Theory and Decision, 31 (1991), 49–73. [9]. T.L. Saaty. The analytic hierarchy process. McGraw-Hill, New York 1980. [10]. S.J. Chen, and C.L. Hwang. Fuzzy multiple attribute decision making. Springer, Berlin, Heidelberg 1992. [11]. Y.J. Wang, and H.S. Lee. Generalizing TOPSIS for fuzzy multiple-criteria group decision-making. Computers and Mathematics with Applications, 53 (2007), 1762–1772. [12]. S Pramanik, K Mondal. Weighted fuzzy similarity measure based on tangent function and its application to medical diagnosis. International Journal of Innovative Research in Science, Engineering and Technology, 4(2) (2015), 158-164. [13]. T.Y. Chen. An ELECTRE-based outranking method for multiple criteria group decision making using interval type-2 fuzzy sets. Information Sciences, 263 (2014), 1–21.
[14]. W. Wang, X. Liu, and Y. Qin. Multi-attribute group decision making models under interval type-2 fuzzy environment. Knowledge-Based Systems, 30 (2012), 121–128. [15]. G.W. Wei. Maximizing deviation method for multiple attribute decision making in intuitionistic fuzzy setting. Knowledge-Based Systems, 21 (2008), 833–836. [16]. G.W. Wei. Gray relational analysis method for intuitionistic fuzzy multiple attribute decision making. Expert Systems with Applications, 38 (2011), 11671–11677. [17]. S. Pramanik, and D. Mukhopadhyaya. Grey relational analysis based intuitionistic fuzzy multi-criteria group decisionmaking approach for teacher selection in higher education. International Journal of Computer Applications, 34 (2011), 21–29. [18]. K. H. Chang, Y. C. Chang, and I. T. Tsai. Enhancing FMEA assessment by integrating grey relational analysis and the decision making trial and evaluation laboratory approach. Engineering Failure Analysis, 31((2013)), 211-224. [19]. K. Mondal and S. Pramanik. Intuitionistic fuzzy multi criteria group decision making approach to quality-brick selection problem. Journal of Applied Quantitative Methods 9 (2) (2014), 35–50. [20]. P. P. Dey, S. Pramanik, and B. C. Giri. Multi-criteria group decision making in intuitionistic fuzzy environment based on grey relational analysis for weaver selection in Khadi institution. Journal of Applied and Quantitative Methods, 10 (4) (2015), 1-14. [21]. K. Mondal and S. Pramanik. Intuitionistic fuzzy similarity measure based on tangent function and its application to multi-attribute decision making. Global Journal of Advanced Research, 2(2) (2015), 464-471. [22]. J. Ren. Sustainability prioritization of energy storage technologies for promoting the development of renewable energy: A novel intuitionistic fuzzy combinative distance-based assessment approach. Renewable Energy, 121 (2018), 666676. [23]. C.T. Chen. Extensions of the TOPSIS for group decisionmaking under fuzzy environment. Fuzzy Sets and Systems, 114 (2000), 1–9. [24]. F.E. Boran, S. Genç, M. Kurt, and D. Akay. A multi-criteria intuitionistic fuzzy group decision making for supplier selection with TOPSIS method. Expert Systems with Applications, 36 (2009), 11363–11368. [25]. X. Zhao. TOPSIS method for interval-valued intuitionistic fuzzy multiple attribute decision making and its application to teaching quality evaluation. Journal of Intelligent and Fuzzy Systems, 26 (2014), 3049–3055. [26]. H. Wang, F. Smarandache, R. Sunderraman, and Y.Q. Zhang. Single valued neutrosophic sets. Multi-space and Multi-structure, 4 (2010), 410–413. [27]. P. Biswas, S. Pramanik, B.C. Giri. Entropy based grey relational analysis method for multi-attribute decision making under single valued neutrosophic assessments. Neutrosophic Sets and Systems, 2 (2014), 102–110. [28]. P. Biswas, S. Pramanik, and B.C. Giri. A new methodology for neutrosophic multi-attribute decision making with unknown weight information. Neutrosophic Sets and Systems, 3 (2014), 42–52.
Pranab Biswas, Surapati Pramanik, and Bibhas C. Giri: TOPSIS strategy for MADM with trapezoidal neutrosophic numbers
38
Neutrosophic Sets and Systems, Vol. 19, 2018
[29]. D. Stanujkic, F. Smarandache, E.K. Zavadskas, and D. Karabasevic. Multiple criteria evaluation model based on the single valued neutrosophic set. Neutrosophic Sets and Systems, 14 (2016), 3-6. [30]. P. Biswas, S. Pramanik, and B.C. Giri. Cosine similarity measure based multi-attribute decision-making with trapezoidal fuzzy neutrosophic numbers. Neutrosophic Sets and Systems, 8 (2015), 47–57. [31]. P. Biswas, S. Pramanik, and B.C. Giri. Aggregation of triangular fuzzy neutrosophic set information and its application to multi-attribute decision making. Neutrosophic Sets and Systems, 12 (2016), 20–40. [32]. R. Sahin, P. Liu. Maximizing deviation method for neutrosophic multiple attribute decision making with incomplete weight information. Neural Computing and Applications, 27(7) (2016), 2017-2029. [33]. Y. Ye. Multicriteria decision-making method using the correlation coefficient under single-valued neutrosophic environment. International Journal of General Systems, 42 (2013), 386–394. [34]. K. Mondal, and S, Pramanik. Neutrosophic tangent similarity measure and its application to multiple attribute decision making. Neutrosophic Sets and Systems, 9 (2015), 85–92. [35]. R. Sahin. Cross-entropy measure on interval neutrosophic sets and its applications in multi criteria decision making. Neural Computing and Applications, 28 (2017), 1177- 1187. [36]. R. Liang, J. Wang, and H. Zhang. A multi-criteria decisionmaking method based on single-valued trapezoidal neutrosophic preference relations with complete weight information. Neural Computing and Applications, (2017), doi: 10.1007/s00521-017-2925-8. [37]. S. Pramanik, S. Dalapati, S. Alam, F. Smarandache, and T.K. Roy. NS-cross entropy-based MAGDM under single-valued neutrosophic set environment. Information (Switzerland), 9 (2018), doi:10.3390/info9020037. [38]. H. Wang, F. Smarandache, R. Sunderraman, and Y.Q. Zhang, Interval neutrosophic sets and logic: Theory and applications in computing. Hexis, Arizona, 2005. [39]. P. Biswas, S. Pramanik, and B.C. Giri. TOPSIS method for multi-attribute group decision-making under single-valued neutrosophic environment. Neural Computing and Applications, 27 (2016), 727–737. [40]. R. Sahin, R. Sahin, and M. Yigider. A multi-criteria neutrosophic group decision making method based TOPSIS for supplier selection. Applied Mathematics and Information Sciences, 10(2016), doi: 10.18576/amis/100525. [41]. Z. Zhang, and C. Wu. A novel method for single-valued neutrosophic multi-criteria decision making with incomplete weight information. Neutrosophic Sets and Systems, 4 (2014), 35–49. [42]. J. Ye. An extended TOPSIS method for multiple attribute group decision making based on single valued neutrosophic linguistic numbers. Journal of Intelligent and Fuzzy Systems, 28 (2015), 247–255.
[43]. X. Peng, and J. Dai. Approaches to single-valued neutrosophic MADM based on MABAC, TOPSIS and new similarity measure with score function. Neural Computing and Applications, 29 (10) (2018), 939–954. [44]. S. Pramanik, P.P. Dey, and B.C., Giri, TOPSIS for single valued neutrosophic soft expert set based multi-attribute decision making problems. Neutrosophic Sets and Systems, 10 (2015), 88-95. [45]. K. Mondal S. Pramanik and F. Smarandache, Rough neutrosophic TOPSIS for multi-attribute group decision making Neutrosophic Sets and Systems, 13 (2016),105-117. [46]. S. Pramanik, P.P. Dey, B.C. Giri, & F. Smarandache. An extended TOPSIS for multi-attribute decision making problems with neutrosophic cubic information. Neutrosophic Sets and Systems17, (2017), 20-28. [47]. P. P. Dey, S. Pramanik, and B. C. Giri. TOPSIS for solving multi-attribute decision making problems under bi-polar neutrosophic environment. In F. Smarandache, & S. Pramanik (Eds), New trends in neutrosophic theory and applications (pp. 65-77). Brussels: Pons Editions, 2016. [48]. S. Pramanik, D. Banerjee, and B. C. Giri. TOPSIS approach for multi attribute group decision making in refined neutrosophic environment. In F. Smarandache, & S. Pramanik (Eds.), New trends in neutrosophic theory and applications (pp. 79-91). Brussels: Pons Editions, 2016. [49]. P. P. Dey, S. Pramanik, and B. C. Giri. Generalized neutrosophic soft multi-attribute group decision making based on TOPSIS. Critical Review 11, 41-55. [50]. J. Ye. Some weighted aggregation operators of trapezoidal neutrosophic numbers and their multiple attribute decision making method. Informatica, 28 (2017), 387–402. [51]. Y. Subas. Neutrosophic numbers and their application to multi-attribute decision making problems (in Turkish). (Master Thesis, Kilis 7 Aralik University, Graduate School of Natural and Applied Science) (2015). [52]. I. Deli, and Y. Şubaş. A ranking method of single valued neutrosophic numbers and its applications to multi-attribute decision making problems. International Journal of Machine Learning and Cybernetics, 8 (2017), 1309–1322. [53]. P. Biswas, S. Pramanik, and B.C. Giri. Value and ambiguity index based ranking method of single-valued trapezoidal neutrosophic numbers and its application to multi-attribute decision making. Neutrosophic Sets and Systems, 12 (2016), 127–138. [54]. D. Dubois, and H. Prade. Ranking fuzzy numbers in the setting of possibility theory. Information Sciences, 30(3) (1983), 183-224. [55]. S. Heilpern. The expected value of a fuzzy number. Fuzzy Sets and Systems, 47(1) (1992), 81-86. [56]. K.S. Park, and S.H. Kim. Tools for interactive multi-attribute decision making with incompletely identified information. European Journal of Operational Research, 98 (1997), 111– 123.
Pranab Biswas, Surapati Pramanik, and Bibhas C. Giri: TOPSIS strategy for MADM with trapezoidal neutrosophic numbers
39
Neutrosophic Sets and Systems, Vol. 19, 2018
[57]. S.H. Kim, and B.S. Ahn. Interactive group decision making procedure under incomplete information. European Journal of Operational Research, 116 (1999), 498–507. [58]. K.S. Park. Mathematical programming models for characterizing dominance and potential optimality when multi-criteria alternative values and weights are simultaneously incomplete. IEEE Transactions on Systems, Man, and CyberneticsPart a: Systems and Humans, 34 (2004), 601–614. [59]. Z.S. Xu, and J. Chen. An interactive method for fuzzy multiple attribute group decision making. Information Sciences, 177 (2007), 248–263. [60]. Y. Wang. Using the method of maximizing deviation to make decision for multiindices. Journal of Systems Engineering and Electronics, 8 (1997), 21–26. [61]. S. Pramanik, S, Dalapati, and T. K. Roy. Logistics center location selection approach based on neutrosophic multi-criteria decision making, In F. Smarandache, & S. Pramanik (Eds.), New Trends in Neutrosophic Theory and Application. Pons Editions, Brussels, 2016, 161-174. [62]. S. Pramanik, S, Dalapati, and T. K. Roy. Neutrosophic multiattribute group decision making strategy for logistic center location selection. In F. Smarandache, M. A. Basset & V. Chang (Eds.), Neutrosophic Operational Research, Vol. III. Pons Asbl, Brussels, 2018, 13-32. [63]. P. P. Dey, S. Pramanik, and B. C. Giri. Multi-criteria group decision making in intuitionistic fuzzy environment based on grey relational analysis for weaver selection in khadi institution. Journal of Applied and Quantitative Methods, 10 (4) (2015), 1-14.
[64]. P. P. Dey, S. Pramanik, and B. C. Giri. An extended grey relational analysis based interval neutrosophic multi-attribute decision making for weaver selection. Journal of New Theory, (9) (2015), 82-93. [65]. K. Mondal, S. Pramanik, and F. Smarandache. Role of neutrosophic logic in data mining. In F. Smarandache, & S. Pramanik (Eds.), New Trends in Neutrosophic Theory and Application. Pons Editions, Brussels, 2016, 15-23. [66]. K. Mondal, and S. Pramanik. Neutrosophic decision making model of school choice. Neutrosophic Sets and Systems, 7 (2015), 62-68. [67]. K. Mondal, and S. Pramanik. Multi-criteria group decision making approach for teacher recruitment in higher education under simplified neutrosophic environment. Neutrosophic Sets and Systems, 6 (2014), 28-34. [68]. K. Mondal, S. Pramanik, and F. Smarandache. Intuitionistic fuzzy multi-criteria group decision making approach to quality-brick selection problem. Journal of Applied and Quantitative Methods, 9(2) (2014), 35-50. [69]. K. Mondal, and S. Pramanik. Neutrosophic decision making model for clay-brick selection in construction field based on grey relational analysis. Neutrosophic Sets and Systems, 9 (2015), 72-79. [15] Abdel-Basset, M.; Mohamed, M.; Smarandache, Flor. An Extension of Neutrosophic AHP–SWOT Analysis for Strategic Planning and Decision-Making. Symmetry 2018, 10, 116.
Received : January 22, 2018. Accepted : March 13, 2018.
Pranab Biswas, Surapati Pramanik, and Bibhas C. Giri: TOPSIS strategy for MADM with trapezoidal neutrosophic numbers