Some hybrid weighted aggregation operators under neutrosophic set environment by Ioan Degău

Some hybrid weighted aggregation operators under neutrosophic set environment and their applications to multi criteria decision making Harish Garg Âˇ Nancy

Abstract Neutrosophic sets (NS) contain the three ranges: truth, indeterminacy, and falsity membership degrees, and are very useful for describing and handling the uncertainties in the real life problem. The aggregation of the neutrosophic sets is one of the important concepts to aggregate the uncertain data. In this paper, some new hybrid aggregation operators based on arithmetic and geometric aggregation operators have been developed to aggregate the information under the single-valued neutrosophic (SVN) set environment. Further, we establish some of its basic properties. Then, we extend these operators to the interval neutrosophic set (INS) environment. Further, an approach based on these operators for multi-criteria decision-making (MCDM) problems is explored under SVN/INS environment. Moreover, to demonstrate its practicality and effectiveness, numerical examples have been presented under both environments. Finally, a comparison analysis has been made with some other existing methods to analyze the superiority of proposed work. Keywords: Neutrosophic Set; single-valued neutrosophic sets; interval neutrosophic set; aggregation operators.

1 Introduction During the decision-making process, it is difficult for the person to get the sufficient and accurate data for real decision-making owing to the vagueness on the uncertainties. To address this issue, Zadeh [1] introduced the concept of the fuzzy set (FS) and since then it has been widely used in many real fields. Further, different types of the FSs have been developed and investigated, namely intuitionistic fuzzy set [2], interval-valued intuitionistic fuzzy set [3], Pythagorean fuzzy set [4], neutrosophic set [5]. In the past few decades, under these environments, researchers have gained great attention and been successfully applied to many practical areas such as decision making, pattern recognition, medical diagnosis, and clustering analysis [6, 7, 8, 9, 10, 11, 12]. Among these FSs, neutrosophic set (NS) seems to be more reasonable and acceptable. In NSs, the uncertain information is represented as a tuple of membership, non-membership and indeterminacy degrees, all are independent and are real standard or non-standard subsets of ]0âˆ’ , 1+ [. Thus, it will be difficult to apply in real scientific and engineering areas. Therefore, Wang et al. [13, 14] proposed the concepts of an interval neutrosophic set (INS) and a single-valued neutrosophic (SVN) set (SVNS), which are the classes of the NSs. Under these environments, it is necessary to aggregate the different SVNSs or INSs and hence their corresponding weighted operators and the ordered weighted operators have been applied to it. For instance, Ye [15] developed the singlevalued neutrosophic weighted averaging (SVNWA) operator and single-valued neutrosophic weighted geometric (SVNWG) operator. Peng et al. [16] defined the operations of SVNSs and based on it some weighted and ordered weighted averaging and geometric aggregation operators, namely single valued neutrosophic ordered weighted average (SVNOWA) and single valued neutrosophic ordered weighted geometric (SVNOWG) operators have been developed. Liu et al. [17] developed some generalized neutrosophic aggregation operators based Harish Garg Âˇ Nancy School of Mathematics, Thapar University Patiala Patiala 147004, Punjab, India E-mail: harishg58iitr@gmail.com URL: http://sites.google.com/site/harishg58iitr/

H.Garg and Nancy

on Hamacher operations and named as single-valued neutrosophic Hamacher weighted averaging (SVNHWA), single-valued neutrosophic Hamacher ordered weighted averaging (SVNHOWA), single-valued neutrosophic Hamacher weighted geometric(SVNHWG) and single-valued neutrosophic Hamacher ordered weighted geometric (SVNHOWG). Zhang et al. [18] proposed an aggregation operator under the INSs and called as interval neutrosophic number weighted averaging (INNWA) and interval neutrosophic number weighted geometric (INNWG) operators. After that, Aiwu et al. [19] proposed the generalized weighted aggregation order under the INS environment. Recently, Nancy and Garg [20] proposed the weighted averaging and geometric aggregation operator based on the Frank norm operators and called as single-valued neutrosophic Frank weighted averaging and geometric operators denoted by SVNFWA and SVNFWG respectively. Apart from that, some other authors have worked under the NS environment and developed their corresponding aggregation operators [21, 22, 23, 24, 25, 26, 27] and their corresponding references. Effective aggregation is one of the most important research areas in the field of decision-making. Aggregation, which usually, involves mathematical operators, is not just an average; rather, it represents a more general notion. The results of the aggregation operator are meaningful if the aggregated value given by it is unbiased, that is, it should never tend to one or some number(s)(among to be combined) whose weight is on the higher side and doesn’t tend to maximum or the minimum arguments. But, the aggregated value of existing weighted averaging and geometric aggregation operators sometimes tends towards the maximum arguments or the argument with higher importance respectively. Hence, they give unrealistic results in some situations. So, the main objective of the manuscript is to introduce some new aggregation operators which have characteristics of both the averaging and geometric operators. In order to achieve this objective, we propose the hybrid weighted arithmetic and geometric aggregation operators under the neutrosophic set environment. These aggregation operators are called as hybrid SVN weighted averaging and geometric and hybrid SVN ordered weighted averaging and geometric. Furthermore, this version of operators give us the averaging and geometric operators as special cases which make its formulation robust. Some of its desirable properties have also been investigated. Later on, these aggregation operators have been extended to the interval neutrosophic set(INS) environment. Finally, an illustrative example of the decision-making problem has been given to show the developed method. The rest of the paper is organized as follows. In section 2, we introduce the neutrosophic set theory along with limitations of weighted averaging and geometric operators. Section 3 proposes the hybrid aggregation operators under SVN environment and also desirable properties of them are investigated. In section 4, we extend these proposed aggregation operators from the SVN to INS domain. Section 5 illustrates the proposed approach with a numerical example and also shows it flexibility by assigning different values to the parameter. Also, the comparative study validates the proposed approach. Section 6 ends up with concluding remarks.

2 Preliminaries In this section, some basic concepts about the SVNSs and INSs are briefly presented over the universal set X .

2.1 Neutrosophic set theory Definition 1 [5] A neutrosophic set (NS) β consists of three independent degrees namely truth (ζβ ), indeterminacy (κβ ), and falsity (ϕβ ) which are defined as β = {hx, ζβ (x), κβ (x), ϕβ (x) | x ∈ Xi}

where ζβ (x), κβ (x), ϕβ (x) ∈]0− , 1+ [ such that 0− ≤ ζβ (x), κβ (x), ϕβ (x) ≤ 3+ . Definition 2 [14]A single-valued neutrosophic set (SVNS) β in X is defined as β = {hx, ζβ (x), κβ (x), ϕβ (x) | x ∈ Xi}

where ζβ (x), κβ (x), ϕβ (x) ∈ [0, 1] such that 0 ≤ ζβ (x), κβ (x), ϕβ (x) ≤ 3. A SVNS is an instance of a NS. For convenience, we denote these pairs as β = hζβ , κβ , ϕβ i, throughout this article, and called as single-valued neutrosophic number (SVNN). Definition 3 An order relation, based on S and H functions, between two SVNNs β and γ is stated as[14], if S (β ) > S (γ ) then β γ and if S (β ) = S (γ ) and H (β ) > H (γ ) then β γ , if H (β ) = H (γ ) then β = γ , where S (β ) = ζβ − κβ − ϕβ and H (β ) = ζβ + κβ + ϕβ .

Some hybrid weighted aggregation operators under neutrosophic set environment

Definition 4 Let β1 = hζ1 , κ1 , ϕ1 i and β2 = hζ2 , κ2 , ϕ2 i be two SVNNs. Then the operational laws between them are defined as follows [16] (i) (ii) (iii) (iv)

β1 ⊕ β2 = hζ1 + ζ2 − ζ1 ζ2 , κ1 κ2 , ϕ1 ϕ2 i β1 ⊗ β2 = hζ1 ζ2 , κ1 + κ2 − κ1 κ2 , ϕ1 + ϕ2 − ϕ1 ϕ2 i λβ1 = h1 − (1 − ζ1 )λ , κ1 λ , ϕ1 λ i; λ > 0 β1 λ = hζ1 λ , 1 − (1 − κ1 )λ , 1 − (1 − ϕ1 )λ i; λ > 0

In order to aggregate the different SVNNs βj = hζj , κj , ϕj i for j = 1, 2, . . . , n corresponding to its weight n P vector ω = (ω1 , ω2 , . . . , ωn )T such that ωj > 0 and ωj = 1, Peng et al. [16] gave the different weighted j =1

geometric and averaging aggregation operators which are summarized as follows: (a) SVNWA operator [16] n Y

* 1−

SVNWA(β1 , β2 , . . . , βn ) =

ωj

(1 − ζj ) ,

j =1

n Y

ωj

(κ j ) ,

j =1

n Y

+ (ϕj )

ωj

(1)

j =1

(b) SVNWG operator [16] n Y

* SVNWG(β1 , β2 , . . . , βn ) =

ωj

(ζj ) , 1 −

j =1

n Y

ωj

(1 − κj ) , 1 −

j =1

n Y

+ (1 − ϕj )

ωj

(2)

j =1

SVNOWA(β1 , β2 , . . . , βn ) =

n Y

ωj

(1 − ζξ(j ) ) ,

n Y

ωj

( κ ξ (j ) ) ,

+ ( ϕ ξ (j ) )

ωj

(3)

j =1

n Y

(d) SVNOWG operator [16] * SVNOWG(β1 , β2 , . . . , βn ) =

n Y

ωj

(ζξ(j ) ) , 1 −

j =1

n Y

ωj

(1 − κξ(j ) ) , 1 −

j =1

n Y

+ (1 − ϕξ(j ) )

ωj

(4)

j =1

where ξ is a permutation of (1, 2, . . . , n) such that βξ(j−1) ≥ βξ(j ) for j = 2, . . . , n.

2.2 Shortcomings of SVNWA, SVNWG, SVNOWA, SVNOWG operators The existing aggregation operators given in Eqs. (1)-(4) are the basic operators used to aggregate the SVNNs but these operators have some shortcomings which are discussed as: Let β1 = h0.0001, 0, 0i and β2 = h1, 0, 0i be two SVNNs and ω = (0.9, 0.1)T be the corresponding weight vector. By utilizing the Eqs. (1)-(4), we get SVNWA(β1 , β2 ) = h1, 0, 0i, SVNWG(β1 , β2 ) = h0.002, 0, 0i, SVNOWA(β1 , β2 ) = h1, 0, 0i and SVNOWG(β1 , β2 ) = h0.5012, 0, 0i. On the other hand, if we take ω = (0.1, 0.9) as weight vector , then by using Eqs. (1)-(4), we get SVNWA(β1 , β2 ) = h1, 0, 0i, SVNWG(β1 , β2 ) = h0.5012, 0, 0i, SVNOWA(β1 , β2 ) = h1, 0, 0i and SVNOWG(β1 , β2 ) = h0.002, 0, 0i. From the above cases, we notice that the aggregated values of SVNWA and SVNOWA operators tend to the maximum argument and the aggregated values of the SVNWG and SVNOWG operator may tend to the value having maximum weight. So, it is concluded from these results that the existing operators may not give the reasonable results and hence there is need to improve the existing aggregation operators so that we can overcome these flaws.

3 Aggregation operators with single-valued neutrosophic information Let Ω be the collection of all SVNNs. In this section, we have proposed some hybrid aggregation operators namely, Hybrid SVN weighted averaging and geometric (H-SVNWAG) and Hybrid SVN ordered weighted averaging and geometric (H-SVNOWAG) aggregation operators.

H.Garg and Nancy

3.1 Hybrid single-valued neutrosophic weighted arithmetic and geometric aggregation operator Definition 5 Let βj = hζj , κj , ϕj i(j = 1, 2, . . . , n) be SVNNs. A H-SVNWAG operator is a mapping H-SVNWAG : Ω n → Ω and is defined as

 H-SVNWAG(β1 , β2 , . . . , βn ) = 

n M

λ

 O

ωj βj 

n O



1−λ ω

βj j 

(5)

j =1

where λ ∈ [0, 1] be a real number and ωj is the standardized weight vector of βj ; (j = 1, 2, . . . , n).

Theorem 1 The aggregated value by using H-SVNWAG operator for a collection of SVNNs βj ; (j = 1, 2, . . . , n) is given in Eq. (5) and still a SVNN.

H-SVNWAG(β1 , β2 , . . . , βn ) 1−λ λ  1−λ  λ  * n n n n Y Y Y Y , (1 − κj )ωj  , 1 − 1 − κj ωj   ζj ωj  = 1 − (1 − ζj )ωj   n Y

1 − 1 −

λ  ϕj ωj  

j =1

n Y

j =1



1−λ + (1 − ϕj )ωj 

(6)

j =1

Proof For SVNNs βj and a real number λ ∈ [0, 1], we have



n M



λ

ωj βj  =

1−

j =1

n Y

ωj

(1 − ζj ) ,

j =1

n Y

ωj

(κ j ) ,

n Y

+!λ (ϕ j )

ωj

j =1

λ + λ  λ  * n n n Y Y Y (1 − ζj )ωj  , 1 − 1 − (κj )ωj  , 1 − 1 − (ϕj )ωj  = 1 −

(7)

j =1

and



n Y



1−λ ω βj j 

n Y

* =

j =1

ωj

(ζj ) , 1 −

j =1

* =



n Y

(ζj )

(1 − κj ) , 1 −

j =1

1−λ

j =1

ωj

ωj 

n Y

+!1−λ (1 − ϕj )

ωj

j =1

 ,1 − 

n Y

1−λ ωj 

(1 − κj )

j =1

 ,1 − 

n Y j =1

1−λ + (1 − ϕj )

ωj 

(8)

Some hybrid weighted aggregation operators under neutrosophic set environment

Therefore, based on these and by Definition 5, we have H-SVNWAG(β1 , β2 , . . . , βn ) * !λ n Y = 1− (1 − ζj )ωj ,1 −

n Y

1−

j=1

!1−λ ωj

n Y

,1 −

(ζj )

1−

n Y

!λ (1 − ζj )

n Y

ωj

n Y

1−

ωj

(1 − κj )

+1 −

1−

n Y

,1 −

(ζj )

!λ !

n Y

1−

(1 − ϕj )ωj

− n Y

1−

(ζj )ωj

1−

−

!λ

n Y

+1−

1−

n Y

(ϕj )ωj !λ

n Y

1−

!1−λ !+ (1 − ϕj )ωj

j=1

!λ

n Y

1−

(κj )ωj

+1−

n Y

1−

!λ

n Y

(κj )ωj

!1−λ −

n Y

1−

!1−λ ! (1 − κj )ωj

j=1

(1 − ϕj )ωj

!1−λ (1 − κj )ωj

j=1

!λ (κj )ωj

n Y

!1−λ (1 − ϕj )ωj

−

j=1

1−

n Y

!λ (ϕj )ωj

j=1

!1−λ !+ (1 − ϕj )ωj

n Y

ωj

(1 − ζj )

!1−λ ζj

ωj

,1 −

1−

j=1

1−

!λ (ϕj )ωj

j=1

1−

(ϕj )ωj

j=1

!λ

j=1 n Y

!λ !

j=1

(ϕj )ωj

n Y

1−

j=1

n Y

,1 −

j=1

n Y

,1 −

!1−λ

(1 − κj )ωj

(1 − κj )

j=1

1−

!1−λ

!1−λ ωj

j=1

!1−λ !

j=1

n Y

+1−

(κj )

j=1

n Y

+ 1−

1−

ωj

j=1

(1 − ζj )ωj

1−

(1 − ϕj ) !λ

n Y

(1 − κj )ωj

!1−λ

j=1

1−

!1−λ + ωj

j=1

(κj )ωj

!λ

− 1−

,1 −

!1−λ ωj

j=1

n Y

j=1

j=1 n Y

!λ + (ϕj )ωj

j=1

!1−λ

j=1

− 1−

n Y

1−

j=1

,1 −

j=1

n Y

!λ (κj )ωj

n Y

!λ ϕj

ωj

j=1

n Y

!λ κj

ωj

j=1 n Y

n Y

!1−λ ωj

(1 − κj )

j=1

!1−λ + ωj

(1 − ϕj )

j=1

Hence, the result. Example 1 Let β1 = h0.9, 0.1, 0.1i, β2 = h0.92, 0.1, 0.05i and β3 = h0.7, 0.1, 0.2i be three SVNNs and their corresponding weight vector is ω = (0.3, 0.5, 0.2)T . Without loss of generality, we assume that λ = 0.5 be a real number. Then, by utilizing the given information, we have H-SVNWAG(β1 , β2 , β3 ) λ  1−λ 1−λ  λ  * 3 3 3 3 Y Y Y Y ω ω ω ω , ζj j  , 1 − 1 − κj j   (1 − κj ) j  = 1 − (1 − ζj ) j    1 − 1 −

3 Y

λ  ϕj ωj  

j =1

* =

j =1

3 Y

j =1

1−λ + (1 − ϕj )ωj 

j =1

1 − (1 − 0.9)0.3 × (1 − 0.92)0.5 × (1 − 0.7)0.2

0.5

(0.90.3 ) × (0.920.5 ) × (0.70.2 )

1−0.5 ,

0.5 0.5 1 − 1 − (0.1)0.3 × (0.1)0.5 × (0.1)0.2 (1 − 0.1)0.3 × (1 − 0.1)0.5 × (1 − 0.1)0.2 , 1 − 1 − (0.1)

0.3

0.5

× (0.05)

× (0.2)

0.2

= h0.8768, 0.1, 0.089i Remark 1 It is evident from the proposed operator that

0.5

0.3

(1 − 0.1)

× (1 − 0.05)

0.5

0.2

× (1 − 0.2)

0.5 +

H.Garg and Nancy

(i) if λ = 1, then H-SVNWAG is reduced to SVNWA [16] operator, and (ii) if λ = 0 then it is reduced to SVNWG [16] operator. Hence, the proposed operator is a more generalized as compared to other existing operator. Furthermore, it has concluded that the proposed H-SVNWAG operator satisfies the properties of idempotency, boundedness and monotonicity for a collection of SVNN βj = hζj , κj , ϕj i; (j = 1, 2, . . . , n), which can be stated as follows: (P1) (Idempotency) If βj = β for all j , we have H-SVNWAG(β1 , β2 , . . . , βn ) = β (P2) (Boundedness) For SVNNs βj ; (j = 1, 2, . . . , n), we have min{βj } ≤ H-SVNWAG(β1 , β2 , . . . , βn ) ≤ max{βj } (P3) (Monotonicity) If βj and γj be two collections of SVNNs such that βj ≤ γj for all j , then H-SVNWAG(β1 , β2 , . . . , βn ) ≤ H-SVNWAG(γ1 , γ2 , . . . , γn ) Since SVNWA and SVNWG operators satisfy these properties so by the definition of H-SVNWAGA operator, it follows directly and hence we omit their proofs. Theorem 2 The existing operators SVNWA, SVNWG and the proposed operator H-SVNWAG satisfy the following inequality for a collection of SVNNs βj (j = 1, 2, . . . , n), SVNWG(β1 , β2 , . . . , βn ) ≤ H-SVNWAG(β1 , β2 , . . . , βn ) ≤ SVNWA(β1 , β2 , . . . , βn )

(9)

Proof Let βj = hζj , κj , ϕj i be SVNNs and ω = (ω1 , ω2 , . . . , ωn )T be their normalized weight vector, then n n n n Q Q Q Q ω ω (1 − ζj )ωj ≤ 1. Therefore, for a real number ζj j , 1 − (1 − ζj )ωj and 0 ≤ ζj j ≤ 1 − we have j =1

j =1

λ ∈ [0, 1] we have



n Y



1−λ ω ζj j 

 ≤ 1 −

j =1

1−λ

n Y

(1 − ζj )ωj 

j =1

which implies that  1 −

n Y

λ  (1 − ζj )ωj  

j =1

Similarly, for

n Q j =1

n Q

κj j ≥ 1 −

ζj j 

≤1−

(1 − κj )ωj and

n Q j =1

 ω

κj j ≤ 1 −  1 −

j =1

n Y

(1 − ζj )ωj

(10)

j =1

n Y

1−λ

n Y

n Q

ϕj j ≥ 1 −

λ 

n Y

ω κj j  

j =1

(1 − ϕj )ωj , we have

j =1

1−λ (1 − κj )ωj 

(11)

j =1

and n Y j =1

 ω ϕj j

≤ 1 − 1 −

n Y j =1

λ  ω ϕj j 

n Y

 j =1

1−λ (1 − ϕj )

ωj 

(12)

Some hybrid weighted aggregation operators under neutrosophic set environment

Thus, by using Eqs. (10), (11), (12) and the definition of score function, we get

S (SVNWA(β1 , β2 , . . . , βn )) = 1 −

n Y

(1 − ζj )ωj −

j =1

n Y

κj j −

n Y

ϕj j

j =1

  

1−λ    ω ω ≥ 1 − (1 − ζj )ωj   ζj j  − 1 − 1 − κj j   (1 − κj )ωj      j =1 j =1 j =1 j =1    λ  1−λ   n n   Y Y ωj   ωj   − 1− 1− ϕj (1 − ϕj )     j =1 j =1 

n Y

λ 

n Y

1−λ



n Y

λ 

n Y

= S (H-SVNWAG(β1 , β2 , . . . , βn ))

Hence, H-SVNWAG(β1 , β2 , . . . , βn ) ≤ SVNWA(β1 , β2 , . . . , βn ). Similarly, we can obtain that H-SVNWAG(β1 , β2 , . . . , βn ) ≥ SVNWG(β1 , β2 , . . . , βn ). Thus, we get the required proof. Example 2 Let β1 = h0.9, 0.1, 0.1i, β2 = h0.92, 0.1, 0.05i and β3 = h0.7, 0.1, 0.2i be three SVNNs and their corresponding weight vector is ω = (0.3, 0.5, 0.2)T as given in Example 1. Then by using Eqs. (1), (2) and (6), we get SVNWA(β1 , β2 , β3 ) = h0.9426, 0.1, 0.08i, H-SVNWAG(β1 , β2 , β3 ) = h0.8768, 0.1, 0.089i and SVNWG(β1 , β2 , β3 ) = h0.865, 0.1, 0.0968i. Therefore, based on the score functions, we get SVNWG(β1 , β2 , β3 ) ≤ H-SVNWAG(β1 , β2 , β3 ) ≤ SVNWA(β1 , β2 , β3 ) and it validates the theorem 2.

3.2 Hybrid single-valued neutrosophic ordered weighted arithmetic and geometric aggregation operator Definition 6 A H-SVNOWAG operator is a mapping H-SVNOWAG(β1 , β2 , . . . , βn ) : Ω n → Ω that has an n P wj = 1 and is defined as associated positional weight vector w = (w1 , w2 , . . . , wn ), such that wj > 0 and j =1

 H-SVNOWAG(β1 , β2 , . . . , βn ) = 

n M

λ wj βξ(j ) 

 1−λ n O w j   β ξ (j )

j =1

(13)

j =1

where ξ is a permutation of (1, 2, . . . , n) such that ξ (j − 1) > ξ (j ) for j = 2, 3, . . . , n and λ be any real number in [0, 1]. Theorem 3 For the collection of SVNNs βj = hζj , κj , ϕj i, (j = 1, 2, . . . , n), the aggregated value by using H-SVNOWAG is still a SVNN and is given by

H-SVNOWAG(β1 , β2 , . . . , βn ) λ  1−λ  λ  1−λ * n n n n Y Y Y Y w w = 1 − (1 − ζξ(j ) )wj   ζξ(jj )  , 1 − 1 − κξ(jj )   (1 − κξ(j ) )wj  , j =1

 1 − 1 −

j =1 n Y j =1

λ  w ϕξ(jj )  

n Y

j =1

1−λ + (1 − ϕξ(j ) )wj 

j =1

(14)

H.Garg and Nancy

Proof Based on the SVNOWA and SVNOWG operators and the operational laws, we have  H-SVNOWAG(β1 , β2 , . . . , βn ) = 

n M

λ

 O

wj βξ(j) 



j=1

n Y

1−

(1 − ζξ(j) )wj ,

n Y

(1 − ϕξ(j) )wj

1−λ wj  βξ(j)

j=1 n Y

(κξ(j) )wj ,

(ϕξ(j) )wj

!λ

n DY

j=1

1−

n Y

n O

(ζξ(j) )wj , 1 −

n Y

(1 − κξ(j) )wj ,

j=1

!1−λ

j=1

λ  λ  λ + * n n n Y Y Y wj  wj  wj     = 1− (1 − ζξ(j) ) ,1 − 1 − (κξ(j) ) ,1 − 1 − (ϕξ(j) ) j=1

j=1

1−λ  1−λ  1−λ + * n n n O Y Y Y  (ζξ(j) )wj  ,1 −  (1 − κξ(j) )wj  ,1 −  (1 − ϕξ(j) )wj  j=1

j=1

λ  λ  λ + * n n n Y Y Y w w w j j j = 1 − (1 − ζξ(j) )  , 1 − 1 − (κξ(j) )  , 1 − 1 − (ϕξ(j) )  j=1

j=1

1−λ  1−λ  1−λ + * n n n O Y Y Y wj  wj  wj     (ζξ(j) ) ,1 − (1 − κξ(j) ) ,1 − (1 − ϕξ(j) ) j=1

j=1

(15)

* 1 −

n Y

λ  wj 

(1 − ζξ(j) )

n Y



j=1

1−λ wj 

n Y

− 1 − 1 −

(ζξ(j) )

n Y

λ wj 

(κξ(j) )

 +1−

j=1

λ !



wj 

1−

(κξ(j) )

j=1

+1 − 

n Y

, 1 − 1 −

j=1



n Y

wj 



1−λ (1 − ϕξ(j) )

, 1 − 1 −

wj 

(ϕξ(j) )

j=1

 −

(1 − κξ(j) ) λ

n Y

j=1

wj 

1−λ wj 

j=1

1−λ ! (1 − κξ(j) )

n Y

1 − 1 −

j=1

n Y

λ ! wj 



n Y

1−

(ϕξ(j) )

j=1

1−λ !+ wj 

(1 − ϕξ(j) )

j=1

  λ  1−λ λ 1−λ * n n n n Y Y Y Y = 1 − (1 − ζξ(j) )wj   (ζξ(j) )wj  , 1 − 1 − (κξ(j) )wj  + 1 −  (1 − κξ(j) )wj  j=1

j=1



1−λ

− 1−

n Y

wj 

(1 − κξ(j) )

j=1

 − 1 −

j=1

 1 − 1 −

n Y

− 1 −

n Y

λ (ϕξ(j) )

wj 

 +1−

* 1−

n Y

λ wj 

(ϕξ(j) )

 + 1 −

1−

!λ

n Y

(1 − ζξ(j) )wj

n Y

1−λ wj 

(1 − ϕξ(j) )

n Y

ϕξ(j) wj

n Y

(κξ(j) )

wj 

 −

1−

n Y



1−λ ! wj 

(1 − κξ(j) )

j=1

1−λ wj 

(1 − ϕξ(j) )

j=1

λ  wj 

(ϕξ(j) )

n Y



1−λ !+ wj 

(1 − ϕξ(j) )

j=1

!1−λ ζξ(j) wj

,1 −

j=1

!λ

j=1

+ 1 −

λ 

j=1

j=1 n Y

(κξ(j) )

n Y

j=1

wj 

j=1



j=1



n Y

λ

1−

n Y j=1

!λ κξ(j) wj

n Y

!1−λ (1 − κξ(j) )wj

j=1

!1−λ + (1 − ϕξ(j) )wj

j=1

Hence, the result. Example 3 Consider β1 = h0.4, 0.2, 0.6i, β2 = h0.3, 0.1, 0.4i and β3 = h0.7, 0.2, 0.1i be three SVNNs. Assume that importance of each SVNN is given as w = (0.6, 0.3, 0.1)T and the λ = 0.5. Based on the score function of SVNNs, we get S (β1 ) = −0.4, S (β2 ) = −0.2 and S (β3 ) = 0.4. Thus, S (β3 ) > S (β2 ) > S (β3 )

Some hybrid weighted aggregation operators under neutrosophic set environment

and hence βξ(1) = β3 , βξ(2) = β2 and βξ(3) = β1 . Therefore, by using the Eq. (14), we get H-SVNOWAG(β1 , β2 , β3 ) λ  1−λ  λ  1−λ * 3 3 3 3 Y Y Y Y w w = 1 − (1 − ζξ(j ) )wj   ζξ(jj )  , 1 − 1 − κξ(jj )   (1 − κξ(j ) )wj  , j =1

 1 − 1 −

j =1 3 Y

λ  w ϕξ(jj )  

j =1

3 Y

j =1

1−λ + (1 − ϕξ(j ) )wj 

j =1

1 − (1 − 0.7)0.6 × (1 − 0.3)0.3 × (1 − 0.4)0.1

0.5

× (0.7)0.6 × (0.3)0.3 × (0.4)0.5

0.5

0.5 1 − 1 − (0.2)0.6 × (0.1)0.3 × (0.2)0.1 (1 − 0.2)0.6 × (1 − 0.1)0.3 × (1 − 0.2)0.1 , + 0.5 0.6 0.3 0.1 0.6 0.3 0.1 1 − 1 − (0.1) × (0.4) × (0.6) (1 − 0.1) × (1 − 0.4) × (1 − 0.6) = h0.5482, 0.1669, 0.2244i According to the properties of the SVNOWA and SVNOWG operators , it is clear that the H-SVNOWAG operator also satisfies the properties of idempotency, boundedness and monotonicity and commutativity. (P1) (Idempotency) If βj = β , for all j , then H-SVNOWAG(β1 , β2 , . . . , βn ) = β (P2) (Boundedness) For a SVNN βj , we have min{βj } ≤ H-SVNOWAG(β1 , β2 , . . . , βn ) ≤ max{βj } (P3) (Monotonicity) If βj ≤ γj for all j , then H-SVNOWAG(β1 , β2 , . . . , βn ) ≤ H-SVNOWAG(γ1 , γ2 , . . . , γn ) (P4) (Commutativity)If (β10 , β20 , . . . , βn0 ) be any permutation of (β1 , β2 , . . . , βn ), then H-SVNOWAG(β1 , β2 , . . . , βn ) = H-SVNOWAG(β 0 1 , β 0 2 , . . . , β 0 n ) Theorem 4 The SVNOWA, SVNOWG and proposed operator H-SVNOWAG satisfies the following inequality SVNOWG(β1 , β2 , . . . , βn ) ≤ H-SVNOWAG(β1 , β2 , . . . , βn ) ≤ SVNOWA(β1 , β2 , . . . , βn )

(16)

Proof Proof is similar to Theorem 2 so we omit here. The suitability of aggregated values of the proposed hybrid aggregation operators is discussed in following example. Example 4 Consider the SVNNs β1 = h0.0001, 0, 0i and β2 = h1, 0, 0i as taken in section 2.2. If we consider the weight vector ω = w = (0.9, 0.1)T , then by using the H-SVNWAG and H-SVNOWAG operators given in Eqs. (6) and (14), we get H-SVNWAG(β1 , β2 ) = h0.0447, 0, 0i and H-SVNOWAG(β1 , β2 ) = h0.7079, 0, 0i. Clearly, H-SVNWAG lie between SVNWA(β1 , β2 ) = h1, 0, 0i and SVNWG(β1 , β2 ) = h0.002, 0, 0i and H-SVNOWAG which has value lie between SVNOWA(β1 , β2 ) = h1, 0, 0i and SVNOWG(β1 , β2 ) = h0.5012, 0, 0i. On the other hand, if we utilize the weight vector ω = w = (0.1, 0.9)T , then from Eqs. (6) and (14) we have, H-SVNWAG(β1 , β2 ) = h0.7079, 0, 0i which is between SVNWA(β1 , β2 ) = h1, 0, 0i and SVNWG(β1 , β2 ) = h0.5012, 0, 0i; H-SVNOWAG(β1 , β2 ) = h0.04447, 0, 0i which is lie between SVNOWA(β1 , β2 ) = h1, 0, 0i and SVNOWG(β1 , β2 ) = h0.002, 0, 0i. From the above, we conclude that the proposed hybrid operators are unbiased as they are not tending towards maximum argument shown by SVNWA and SVOWA operators and also not tending towards the maximum weight value as shown by SVNWG and SVNOWG operators. Hence, the proposed operators gives the aggregated values which are much more informative.

H.Garg and Nancy

4 Extensions to the Interval neutrosophic information Wang et al. [13] have introduced the interval neutrosophic set (INS), which is generalization of the SVNS in which the degree of the membership functions corresponding to the set are represented in the form of the intervals rather than exact numbers. Definition 7 An interval neutrosophic set β is defined over X is given as [13] β = {hx, ζ˜β (x), κ ˜β (x), ϕ˜β (x)i | x ∈ X}

where ζ˜β (x), κ ˜β (x), ϕ˜β (x) ⊂ [0, 1] are interval numbers such that 0 ≤ sup(ζ˜β (x))+sup(˜ κβ (x))+sup(ϕ ˜β (x)) ≤ 1 for all x ∈ X . For convenience, let ζ˜β (x) = [a, b], κ ˜β (x) = [c, d] and ϕ˜β (x) = [e, f ] then this pair is often denoted by β = h[a, b], [c, d], [e, f ]i and called an interval-valued neutrosophic number (INN). For any INN β = h[a, b], [c, d], [e, f ]i, a score function S is defined as S (β ) = (a + b − c − d − e − f )/2, S (β ) ∈ [−1, 1] and an accuracy degree H is defined as H (β ) = (a + b + c + d + e + f )/2, H (β ) ∈ [0, 1]. An order relation between these two INNs is defined as : if S (β ) < S (γ ) then β is smaller than γ and if S (β ) = S (γ ) then, if H (β ) < H (γ ) then β is smaller than γ and if H (β ) = H (γ ) then β and γ represent the same information denoted by β = γ . Furthermore, in order to aggregate the different INNs βj = h[aj , bj ], [cj , dj ], [ej , fj ]i, (j = 1, 2, . . . , n), the following aggregation operators have been defined as follows Definition 8 Let βj = h[aj , bj ], [cj , dj ], [ej , fj ]i, (j = 1, 2, . . . , n) be n INNs then (i) an interval neutrosophic weighted average (INWA) operator is defined as [28]

INWA(β1 , β2 , . . . , βn ) =

1−

n Y

1 − aj

ωj

,1 −

n Y

1 − bj

n ωj i h Y

j =1

j =1 n hY

(ej )ωj ,

n Y

(cj )ωj ,

(fj )ωj

(dj )ωj

j =1

n Y

j =1

(ii) an interval neutrosophic weighted geometric (INWG) operator is defined as [28]

INWG(β1 , β2 , . . . , βn ) =

n Dh Y

(aj )ωj ,

j =1

n Y

n n i h Y Y ω ω i 1 − cj j , 1 − 1 − dj j , (bj )ωj , 1 −

j =1

+ n n h Y Y ωj ωj i 1− 1 − ej ,1 − 1 − fj j =1

where ωj be the weight vector of it such that ωj > 0 and

j =1

j =1 ωj

= 1.

In the following, we have proposed some new interval neutrosophic weighted aggregation operators namely, hybrid interval neutrosophic weighted and geometric average (H-INWAG) operator and hybrid interval neutrosophic ordered weighted and geometric average (H-INOWAG)P operator for a collection of the INNs, denoted by Γ , whose weight vector is ω = (ω1 , ω2 , . . . , ωn )T , ωj > 0, n j =1 ωj = 1. Definition 9 For any real number λ ∈ [0, 1] and for the collection of INNs βj = h[aj , bj ], [cj , dj ], [ej , fj ]i, (j = 1, 2, . . . , n), a H-INWAG operator is a mapping H-INWAG : Γ n → Γ and is defined as

Some hybrid weighted aggregation operators under neutrosophic set environment

H-INWAG(β1 , β2 , . . . , βn ) n n λ O O M ω 1−λ βj j = ωj βj j =1

j =1

# *" n n n n 1−λ λ Y 1−λ λ Y Y Y ωj ωj ωj ωj bj , 1− (1 − bj ) aj = 1− (1 − aj ) j =1

j =1

n n n λ 1−λ λ Y Y Y ,1 − 1 − dj ωj (1 − cj )ωj 1− 1− cj ωj

(1 − dj )ωj

j =1

j =1 n Y

1−λ

# "

n Y

, 1− 1−

1− 1−

n Y

ωj

n λ Y

ωj

(1 − fj )

(1 − ej )ωj

1−λ

j =1

ej ωj

n λ Y

1−λ

#+ (17)

j =1

where ωj is the weight of βj ; (j = 1, 2, . . . , n). Theorem 5 The INOWA, INOWG and proposed operator H-INOWAG satisfies the following inequality INWG(β1 , β2 , . . . , βn ) ≤ H-INWAG(β1 , β2 , . . . , βn ) ≤ INWA(β1 , β2 , . . . , βn )

(18)

Proof Proof is similar to Theorem 2, so we omit here. Definition 10 Let βj = h[aj , bj ], [cj , dj ], [ej , fj ]i, j = 1, 2, . . . , n be a collection of INNs. An H-INOWAG n operator of dimension n is a mapping H-INOWA Pn: Γ → Γ , that has an associated positional weight vector T w = (w1 , w2 , . . . , wn ) , such that wj > 0 and j =1 wj = 1, Furthermore, H-INOWAG(β1 , β2 , . . . , βn ) n n M λ O O 1−λ w = wj βξ(j ) βξ(jj ) j =1

j =1

*" # n n n n λ Y 1−λ λ Y 1−λ Y Y wj wj wj wj = 1− (1 − aξ(j ) ) aξ(j ) , 1− (1 − bξ(j ) ) b ξ (j ) , j =1

j =1

n n n λ Y 1−λ λ Y Y c ξ (j ) w j (1 − cξ(j ) )wj ,1 − 1 − dξ(j ) wj 1− 1− j =1

n Y

(1 − dξ(j ) )wj

j =1

1−λ

j =1

n n λ Y 1−λ Y , 1− 1− eξ(j ) wj (1 − eξ(j ) )wj ,

# "

j =1

n n λ Y 1−λ Y fξ(j ) wj (1 − fξ(j ) )wj 1− 1− j =1

j =1

#+ (19)

j =1

where (ξ (1), ξ (2), . . . , ξ (n)) is a permutation of (1, 2, . . . , n) such that βξ(j−1) ≥ βξ(j ) for j = 2, 3, . . . , n. Theorem 6 The INOWA, INOWG and proposed operator H-INOWAG satisfies the following inequality INOWG(β1 , β2 , . . . , βn ) ≤ H-INOWAG(β1 , β2 , . . . , βn ) ≤ INOWA(β1 , β2 , . . . , βn )

(20)

Proof Proof is similar to Theorem 2 so we omit here. 5 Proposed operators based decision-making approach In this section, we have presented a decision-making method for solving MCDM problem by using proposed aggregation operators. A practical example from a field of decision-making has been taken for an illustrative and finally the proposed result has been compared with the existing approaches results under neutrosophic environment for demonstrating the validity of the proposed approach.

H.Garg and Nancy

5.1 Proposed approach Assume that there is a set of m alternatives A = {A1 , A2 , . . . , Am } which are evaluated with respect to the set of n criteria C = {C1 , C2 , . . . , Cn }. Consider that decision maker gives his/her preference values corresponding to each alternative Ai ;(i = 1, 2, . . . , m) with respect to criteria Cj ;(j = 1, 2, . . . , n) in terms of neutrosophic numbers denoted by βij . The collected information has been summarized in terms of neutrosophic decision matrix D and is represented as

C1 A1  β11 A2  β21

.. .

C2 β12 β22

... ... ...

.. .

βm2

...

 .  . .

βm1

Cn β 1n  β 2n 

..   .

βmn

Approach I: Assume that the decision maker/expert gave their preferences towards the alternative in the form of SVNNs βij = hζij , κij , ϕij i, i = 1, 2, . . . , m; j = 1, 2, . . . , n where ζij , ϕij represent the membership and non membership degrees of alternative Ai corresponding to the attribute Cj respectively and κij represent the degree of indeterminacy. Then, in the following, we develop an approach, based on the proposed operators to find the best alternative(s) which involves the following steps. Step 1. Obtain the collective information in terms of the decision matrix D = (βij )m×n . Step 2. Utilize appropriately the H-SVNWAG operator: ri = H-SVNWAG(βi1 , βi2 , . . . , βin )

* n n n n λ Y 1−λ λ Y 1−λ Y Y 1− (1 − ζij )ωj ζij ωj ,1 − 1 − κij ωj (1 − κij )ωj , = j =1

j =1

n n λ Y 1−λ Y ϕij ωj (1 − ϕij )ωj 1− 1− j =1

j =1

or the H-SVNOWAG operator ri = H-SVNOWAG(βi1 , βi2 , . . . , βin )

1−

n Y

(1 − ζiξ(j ) )wj

j =1

n λ Y j =1

1−

n Y

ζiξj(j )

1−λ

,1 − 1 −

n Y

κiξj(j )

j =1

n λ Y 1−λ w 1− ϕiξj(j ) (1 − ϕiξ(j ) )wj j =1 j =1

n λ Y

(1 − κiξ(j ) )wj

1−λ

j =1

to aggregate all the individual SVNNs βij (j = 1, 2, . . . , n) into the collective SVNN ri (i = 1, 2, . . . , m). Step 3. Compute the score value of the aggregated SVNNs ri , (i = 1, 2, . . . , m) and arrange their values in the descending order. Step 4. Rank the alternative according to the score value of final aggregated numbers and hence choose the best one(s). Approach II: If the decision-makers’ are given their preferences towards each alternative Ai (i = 1, 2, . . . , m) in the form of INNs βij = h[aij , bij ], [cij , dij ], [eij , fij ]i, i = 1, 2, . . . , m; j = 1, 2, . . . , n rather than the crisp numbers hζij , κij , ϕij i for i = 1, 2, . . . , m; j = 1, 2, . . . , n then in the following, we develop an approach based on interval operators which involves the following steps. Step 1. Obtain the collective information in terms of the interval-valued decision matrix D = (βij )m×n .

Some hybrid weighted aggregation operators under neutrosophic set environment

Step 2. Utilize appropriately H-INWAG operator ri = H-INWAG(βi1 , βi2 , . . . , βin )

1−

n Y

(1 − aij )

ωj

n λ Y

1−λ

, 1−

j =1

aij

ωj

n Y

(1 − bij )

ωj

n λ Y

bij

ωj

1−λ

j =1

n n n λ 1−λ λ Y Y Y ,1 − 1 − dij ωj (1 − cij )ωj 1− 1− cij ωj j =1

j =1

j =1 n Y

(1 − dij )ωj

1−λ

n n 1−λ λ Y Y , (1 − eij )ωj , 1− 1− eij ωj

# "

j =1

n n 1−λ λ Y Y (1 − fij )ωj 1− 1− fij ωj

j =1

or the H-INOWAG operator ri = H-INOWAG(βi1 , βi2 , . . . , βin )

*" n n n 1−λ λ λ Y Y Y aiξ(j ) wj , 1− (1 − biξ(j ) )wj = 1− (1 − aiξ(j ) )wj j =1

j =1 n Y

biξ(j ) wj

1−λ

# "

j =1

, 1− 1−

1− 1−

n Y

diξ(j )

n λ Y

j =1 n Y

(1 − eiξ(j ) )

j =1

ciξ(j ) wj

n λ Y

j =1

n Y

(1 − ciξ(j ) )wj

j =1

(1 − diξ(j ) )

1−λ

# "

, 1− 1−

j =1

1−λ

n Y

eiξ(j ) wj

j =1

,1 − 1 −

n Y j =1

fiξ(j )

n λ Y

(1 − fiξ(j ) )

1−λ

j =1

to aggregate the preference values into the collected INN ri (i = 1, 2, . . . , m). Step 3. Compute the score value of the aggregated INNs ri , (i = 1, 2, . . . , m) and arrange their values in the descending order. Step 4. Rank the alternative according to the score value of final aggregated numbers and hence choose the best one(s).

5.2 Illustrative example The above proposed operators have been illustrated with a numerical example from the field of decisionmaking. Demonetization is the withdrawal of a particular form of currency from circulation. On 8 November 2016, Government of India announced in a broadcast to the nation that Rs. 500 and Rs. 1,000 currency notes would no longer be recognized legally as currency. This step was taken to crack down the use of illicit and counterfeit cash to fund illegal activity and terrorism. As Government knew that demonetization will effect the Indian economy and may also drop country’s Gross domestic product (GDP) growth. So before the announcement of the demonetization, Government wanted to conduct the survey of effect of this bold move on various sectors of Indian economy to take the further decisions. For doing this, Indian government hired a economist or decision maker who is able to handle this kind of situation and able to crack that which sector (alternative) of Indian economy will be effected by the demonetization. For this, decision maker assume the five important sectors on which our Indian economy depends and were given as: A1 (Agriculture Sector), A2 (Real-Estate Sector ), A3 (Information Technology Sector), A4 (Educational Sector), A5 (Industrial Sector). For evaluation, decision maker considered criterion in the terms ‘how much effect of demonetization on particular sector in linguistic terms’ which are summarized as: C1 (Very low effect), C2 (Low effect), C3 (Regular effect), C4 ( High effect) and C5 (Very high effect). The importance of each criteria Cj ; (i = 1, 2, 3, 4, 5) is taken in the form of weight vector as ω = (0.2, 0.25, 0.15, 0.3, 0.1)T in the decision-making problem and in order to make the decision more pessimistic for future goals then manipulate the aggregation by using the OWA weighted vector w =

H.Garg and Nancy

(0.1, 0.2, 0.2, 0.2, 0.3)T . The procedure to get the most desirable alternative(s) by using the proposed operator is discussed as follows: Step 1. The evaluation of these strategies are taken using the neutrosophic numbers by the decision makers under the above five general characteristics and hence construct the decision matrix as given in Table 1. Insert Table 1 here. Step 2. By utilizing the proposed arithmetic and geometric aggregation operator to aggregate these decision information, their corresponding results are summarized in Table 2 which include the SVNWA, SVNWG, SVNOWA, SVNOWG, SVNFWA, SVNFWG, SVNHWA, SVNHWG, H-SVNWAG and H-SVNOWAG operators. Insert Table 2 here. Step 3. The score values of aggregated SVNNs, shown in Table 2, are summarized in Table 3. Step 4. According to score function of the aggregated values, the ordering of the alternatives is shown in Table 4 in which means “preferred to”. From these results, it has been seen that the best alternative is A2 by all the operators while the different aggregation operators have different ranking strategies which is slightly different. Thus, based on the decision makers preference in terms of their aggregation operators used, the results may leads to the different decisions. On the other hand, if the given five strategies Ai (i = 1, 2, . . . , 5) are to be evaluated using the INNs by the decision makers under the above five characteristics, then the following steps have been executed to find the most desirable alternative(s). Step 1: The rating values of each alternative is measured in the form of INNs and are summarized in Table 5. Insert Table 5 here. Step 2: By using the information provided by the decision maker related to these strategies and the weight vector, we compute, the aggregated values corresponding to it by using INWA, INWG, INOWA, INOWG, INFWA, INFWG, INHWA, INHWG, H-INWAG and H-INOWAG operators. The results corresponding to these are summarized in Table 6. Step 3: Calculate the score of aggregated values by using Definition 7 and summarized in Table 7. Step 4: Based on these score values, we summarized their preference orders in Table 8 and it has been concluded that the best region for investment is A2 .

5.3 Sensitivity Analysis In order to access the impact of the parameter λ on the score values and ranking of the alternative, an investigation has been done by making a change in values of the parameter λ from 0 to 1. The overall score values of the alternative based on the different λ0 s are summarized in the Table 9. From this table, it has been seen that when λ = 0, then the H-SVNWAG/H-INWAG reduces to SVNWG/INWG and H-SVNOWAG/H-INOWAG reduces to SVNOWG/INOWG. Also, it has been highlighted that if λ = 1, then the H-SVNWAG/H-INWAG reduces to SVNWA/INWA and H-SVNOWAG/H-INOWAG reduces to SVNOWA/INOWA. Further, the overall score values of different alternates are increasing as the increase of λ. Also, the ranking order of H-SVNWAG/HINWAG and H-SVNOWAG/H-INOWAG for different values of λ are not identical but the final decision given by them is identical. The decision maker can select the desired value λ according to his preference or practical demand. Thus, the proposed operators are more flexible than the existing operators. Insert Table 9 here

6 Conclusion The main objective of this paper is to present new aggregation operators which are capable to fuse the behavior of the existing averaging and geometric operators. From these existing operators, it has been observed that the aggregated values given by them are either tends towards the maximum arguments values or towards the maximum weight values in some cases and hence they don’t give the unbiased aggregated values. For handling their limitations, the new hybrid neutrosophic weighted average and geometric aggregation operators have

Some hybrid weighted aggregation operators under neutrosophic set environment

been developed which can give the moderate values in the aggregation process. We have also studied some properties of these operators such as idempotency, boundedness, and monotonicity. Moreover, the parameter λ makes the proposed operator more flexible and general than existing operators such as SVNWA/INWA, SVNOWA/INOWA, SVNWG/INWG and SVNOWG/IVNOWA. Further, from the results and their corresponding comparative studies, it has been observed that the decision-making approaches defined in this paper are more stable for solving the MCDM problems. In future, we will extend the proposed work to other complex fields.

References 1. L. A. Zadeh, Fuzzy sets, Information and Control 8 (1965) 338–353. 2. K. T. Atanassov, Intuitionistic fuzzy sets, Fuzzy Sets and Systems 20 (1986) 87 – 96. 3. K. Atanassov, G. Gargov, Interval-valued intuitionistic fuzzy sets, Fuzzy Sets and Systems 31 (1989) 343 – 349. 4. R. R. Yager, A. M. Abbasov, Pythagorean membeship grades, complex numbers and decision making, International Journal of Intelligent Systems 28 (2013) 436 – 452. 5. F. Smarandache, A unifying field in logics. Neutrosophy: neutrosophic probability, set and logic, American Research Press, Rehoboth, 1999. 6. Z. S. Xu, Intuitionistic fuzzy aggregation operators, IEEE Transactions of Fuzzy Systems 15 (2007) 1179 – 1187. 7. H. Garg, Confidence levels based Pythagorean fuzzy aggregation operators and its application to decisionmaking process, Computational and Mathematical Organization Theory (2017) 1–26. URL 10.1007/s10588-017-9242-8 8. H. Garg, Novel intuitionistic fuzzy decision making method based on an improved operation laws and its application, Engineering Applications of Artificial Intelligence 60 (2017) 164 – 174. 9. H. Garg, Generalized intuitionistic fuzzy interactive geometric interaction operators using Einstein t-norm and t-conorm and their application to decision making, Computer and Industrial Engineering 101 (2016) 53 – 69. 10. K. Kumar, H. Garg, TOPSIS method based on the connection number of set pair analysis under intervalvalued intuitionistic fuzzy set environment, Computational and Applied Mathematics (2016) 1–11. URL http://dx.doi.org/10.1007/s40314-016-0402-0 11. H. Garg, A new generalized improved score function of interval-valued intuitionistic fuzzy sets and applications in expert systems, Applied Soft Computing 38 (2016) 988 – 999. 12. H. Garg, A new generalized Pythagorean fuzzy information aggregation using Einstein operations and its application to decision making, International Journal of Intelligent Systems 31 (9) (2016) 886 – 920. 13. H. Wang, F. Smarandache, Y. Q. Zhang, R. Smarandache, Interval neutrosophic sets and logic: theory and applications in computing, Hexis, Phoenix, AZ, 2005. 14. H. Wang, F. Smarandache, Y. Q. Zhang, R. Sunderraman, Single valued neutrosophic sets, Multispace Multistructure 4 (2010) 410 – 413. 15. J. Ye, A multicriteria decision-making method using aggregation operators for simplified neutrosophic sets, Journal of Intelligent and Fuzzy Systems 26 (5) (2014) 2459 – 2466. 16. J. J. Peng, J. Q. Wang, J. Wang, H. Y. Zhang, Z. H. Chen, Simplified neutrosophic sets and their applications in multi-criteria group decision-making problems, International Journal of System Science 47 (10) (2016) 2342 – 2358. 17. P. Liu, Y. Chu, Y. Li, Y. Chen, Some generalized neutrosophic number hamacher aggregation operators and their application to group decision making, International Journal of Fuzzy Systems 16 (2) (2014) 242 – 255. 18. H. Y. Zhang, J. Q. Wang, X. H. Chen, Interval neutrosophic sets and their application in multicriteria decision making problems, The Scientific World Journal Volume 2014 (2014) Article ID 645953, 15 pages. 19. Z. Aiwu, D. Jianguo, G. Hongjun, Interval valued neutrosophic sets and multi-attribute decision-making based on generalized weighted aggregation operator, Journal of Intelligent and Fuzzy Systems 29 (2015) 2697 – 2706. 20. Nancy, H. Garg, Novel single-valued neutrosophic decision making operators under frank norm operations and its application, International Journal for Uncertainty Quantification 6 (4) (2016) 361 – 375. 21. P. Ji, J.-Q. Wang, H.-Y. Zhang, Frank prioritized bonferroni mean operator with single-valued neutrosophic sets and its application in selecting third-party logistics providers, Neural Computing and Applications

22.

23. 24. 25. 26.

27. 28.

H.Garg and Nancy

(2016) 1–25. URL 10.1007/s00521-016-2660-6 Y.-X. Ma, J. Q. Wang, J. Wang, X.-H. Wu, An interval neutrosophic linguistic multi-criteria group decision-making method and its application in selecting medical treatment options, Neural Computing and Applications (2016) 1–21. URL 10.1007/s00521-016-2203-1 Y. Li, P. Liu, Y. Chen, Some single valued neutrosophic number heronian mean operators and their application in multiple attribute group decision making, Informatica 27 (1) (2016) 85 – 110. Nancy, H. Garg, An improved score function for ranking neutrosophic sets and its application to decisionmaking process, International Journal for Uncertainty Quantification 6 (5) (2016) 377 – 385. H. Garg, Nancy, On single-valued neutrosophic entropy of order α, Neutrosophic Sets and Systems 14 (2016) 21 – 28. L. Yang, B. Li, A multi-criteria decision-making method using power aggregation operators for singlevalued neutrosophic sets, International Journal of Database and theory and application 9 (2) (2016) 23 – 32. P. Liu, Y. Wang, Interval neutrosophic prioritized OWA operator and its application to multiple attribute decision making, Journal of Science Science and Complexity 29 (3) (2016) 681 – 697. Z. Zhang, 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.

Table 1 Neutrosophic decision making matrix C1 h0.5, 0.3, 0.4i h0.7, 0.1, 0.3i h0.5, 0.3, 0.4i h0.7, 0.3, 0.2i h0.4, 0.1, 0.3i

A1 A2 A3 A4 A5

C2 h0.5, 0.2, 0.3i h0.7, 0.2, 0.3i h0.6, 0.2, 0.4i h0.7, 0.2, 0.2i h0.5, 0.1, 0.2i

C3 h0.2, 0.2, 0.6i h0.6, 0.3, 0.2i h0.6, 0.1, 0.2i h0.4, 0.5, 0.2i h0.4, 0.1, 0.5i

C4 h0.3, 0.2, 0.4i h0.6, 0.4, 0.2i h0.5, 0.1, 0.3i h0.5, 0.2, 0.2i h0.4, 0.3, 0.6i

C5 h0.3, 0.3, 0.4i h0.7, 0.1, 0.2i h0.6, 0.4, 0.3i h0.4, 0.5, 0.4i h0.3, 0.2, 0.4i

Table 2 Neutrosophic aggregated results by using aggregated operators(λ = 0.5)

A1 A2 A3 A4 A5 A1 A2 A3 A4 A5

SVNWA [16]

SVNWG [16]

h0.3862, 0.2259, 0.3956i h0.6585, 0.2125, 0.2400i h0.5528, 0.1702, 0.3213i h0.5842, 0.2727, 0.2144i h0.4178, 0.1490, 0.3708i SVNOWG [16] h0.3096, 0.2416, 0.4605i h0.6481, 0.2598, 0.2416i h0.5477, 0.2483, 0.3432i h0.4947, 0.3842, 0.2661i h0.3862, 0.1848, 0.4563i

h0.3553, 0.2314, 0.4132i h0.6531, 0.2548, 0.2467i h0.5477, 0.2020, 0.3337i h0.5502, 0.3074, 0.2227i h0.4110, 0.1751, 0.4271i SVNFWA[20] h0.3840, 0.2262, 0.3968i h0.6581, 0.2149, 0.2404i h0.5524, 0.1716, 0.3222i h0.5815, 0.2748, 0.2148i h0.4173, 0.1501, 0.3751i

SVNHWA [17] γ=2 γ=3 h0.3812, 0.2264, 0.3980i h0.3782, 0.2266, 0.3992i h0.6578, 0.2165, 0.2407i h0.6575, 0.2181, 0.2410i h0.5520, 0.1726, 0.3230i h0.5516, 0.1735, 0.3237i h0.5790, 0.2765, 0.2151i h0.5766, 0.2781, 0.2154i h0.4167, 0.1508, 0.3788i h0.4160, 0.1515, 0.3825i SVNFWG[20] H-SVNWAG h0.3576, 0.2311, 0.4117i h0.3704, 0.2286, 0.4045i h0.6535, 0.2525, 0.2463i h0.6558, 0.2340, 0.2434i h0.5481, 0.2002, 0.3329i h0.5502, 0.1862, 0.3275i h0.5528, 0.3047, 0.2220i h0.5669, 0.2903, 0.2185i h0.4115, 0.1738, 0.4230i h0.4144, 0.1622, 0.3996i

SVNOWA [16] h0.3413, 0.2352, 0.4389i h0.6536, 0.2169, 0.2352i h0.5528, 0.2107, 0.3326i h0.5301, 0.3429, 0.2462i h0.4193, 0.1261, 0.3455i H-SVNOWAG h0.3251, 0.2384, 0.4498i h0.6508, 0.2386, 0.2384i h0.5502, 0.2297, 0.3380i h0.5121, 0.3639, 0.2563i h0.3893, 0.1723, 0.4365i

Table 3 Score values of aggregated SVNNs A1 A2 A3 A4 A5

SVNWA [16] -0.2353 0.2060 0.0613 0.0971 -0.1020 SVNFWG [20]

A1 A2 A3 A4 A5

-0.2852 0.1547 0.0150 0.0261 -0.1853

SVNWG [16] SVNOWA[16] -0.2894 -0.3328 0.1516 0.2015 0.0120 0.0094 0.0201 -0.0591 -0.1913 -0.1834 SVNHWA[17] γ=2 γ=3 -0.2432 -0.2476 0.2006 0.1984 0.0564 0.0544 0.0874 0.0831 -0.1129 -0.1180

SVNOWG [16] SVNFWA [20] -0.3924 -0.2390 0.1467 0.2028 -0.0438 0.0886 -0.1556 0.0919 -0.2549 -0.1079 Proposed Operators H-SVNWAG H-SVNOWAG -0.2627 -0.3631 0.1785 0.1738 0.0365 -0.0174 0.0581 -0.1081 -0.1474 -0.2196

Some hybrid weighted aggregation operators under neutrosophic set environment

Table 4 Ordering of the alternatives Existing operators SVNWA [16] SVNOWA [16] SVNWG [16] SVNOWG [16] SVNHWA [17] SVNFWA [20] SVNFWG [20]

A2 A2 A2 A2 A2 A2 A2

A4 A3 A4 A3 A4 A4 A4

Ordering A3 A5 A4 A5 A3 A5 A4 A5 A3 A5 A3 A5 A3 A5

A1 A1 A1 A1 A1 A1 A1

Proposed operators H-SVNWAG H-SVNOWAG

Ordering A2 A3 A4 A5 A1 A2 A4 A3 A5 A1

Table 5 Information about each alternative under five characteristics in form of INNs C1

h[0.4, 0.6], [0.2, 0.3], [0.3, 0.5]i

h[0.45, 0.55], [0.1, 0.2], [0.3, 0.4]i

h[0.2, 0.3], [0.1, 0.2], [0.6, 0.7]i

h[0.2, 0.4], [0.2, 0.3], [0.3, 0.4]i

h[0.2, 0.4], [0.3, 0.4], [0.4, 0.6]i

h[0.7, 0.8], [0.1, 0.2], [0.3, 0.5]i

h[0.65, 0.80], [0.2, 0.3], [0.3, 0.5]i

h[0.5, 0.7], [0.2, 0.3], [0.2, 0.3]i

h[0.6, 0.7], [0.4, 0.5], [0.2, 0.3]i

h[0.6, 0.7], [0.1, 0.3], [0.2, 0.3]i

h[0.5, 0.6], [0.3, 0.4], [0.4, 0.5]i

h[0.60, 0.65], [0.2, 0.3], [0.3, 0.4]i

h[0.4, 0.6], [0.1, 0.2], [0.2, 0.3]i

h[0.4, 0.5], [0.1, 0.3], [0.3, 0.4]i

h[0.6, 0.7], [0.4, 0.5], [0.3, 0.4]i

h[0.6, 0.7], [0.3, 0.5], [0.2, 0.4]i

h[0.70, 0.80], [0.1, 0.2], [0.1, 0.2]i

h[0.4, 0.5], [0.5, 0.6], [0.1, 0.2]i

h[0.3, 0.5], [0.1, 0.2], [0.2, 0.3]i

h[0.3, 0.4], [0.5, 0.6], [0.4, 0.5]i

h[0.4, 0.5], [0.1, 0.2], [0.3, 0.4]i

h[0.50, 0.60], [0.1, 0.2], [0.2, 0.3]i

h[0.3, 0.4], [0.1, 0.2], [0.5, 0.6]i

h[0.3, 0.4], [0.2, 0.3], [0.5, 0.6]i

h[0.2, 0.3], [0.2, 0.4], [0.3, 0.4]i

Table 6 Neutrosophic aggregated results by using aggregated operators(λ = 0.50)

INWA [28]

INWG [28]

h[0.3123, 0.4731], [0.1578, 0.2625], [0.3426, 0.4737]i

h[0.2814, 0.4499], [0.1725, 0.2729], [0.3662, 0.4993]i

h[0.6223, 0.7500], [0.2000, 0.3224], [0.2400, 0.3775]i

h[0.6143, 0.7434], [0.2398, 0.3501], [0.2467, 0.3984]i

h[0.4980, 0.5981], [0.1702, 0.3147], [0.2990, 0.4006]i

h[0.4820, 0.5885], [0.2020, 0.3304], [0.3075, 0.4080]i

h[0.5052, 0.6344], [0.1863, 0.3162], [0.1625, 0.2844]i

h[0.4447, 0.5882], [0.2611, 0.3876], [0.1852, 0.3077]i

h[0.3676, 0.4691], [0.1320, 0.2421], [0.3590, 0.4468]i

h[0.3467, 0.4497], [0.1414, 0.2532], [0.3985, 0.4804]i

INOWA [28]

INOWG[28]

h[0.2725, 0.4456], [0.1835, 0.2896], [0.3757, 0.5283]i

h[0.2491, 0.4228], [0.2038, 0.3043], [0.4024, 0.5541]i

h[0.6043, 0.7344], [0.2000, 0.3358], [0.2259, 0.3497]i

h[0.5970, 0.7286], [0.2398, 0.3587], [0.2314, 0.3672]i

h[0.5170, 0.6155], [0.2107, 0.3478], [0.3140, 0.4156]i

h[0.5030, 0.6063], [0.2483, 0.3667], [0.3226, 0.4231]i

h[0.4424, 0.5649], [0.2786, 0.4161], [0.2000, 0.3280]i

h[0.3973, 0.5242], [0.3621, 0.4851], [0.2398, 0.3613]i

h[0.3169, 0.4182], [0.1414, 0.2670], [0.4119, 0.4571]i

h[0.2961, 0.3995], [0.1515, 0.2855], [0.4395, 0.4819]i

INHWA[17] γ=2

γ=3

h[0.3072, 0.4693], [0.1589, 0.2637], [0.3454, 0.4774]i

h[0.3040, 0.4672], [0.1593, 0.2641], [0.3468, 0.4794]i

h[0.6212, 0.7493], [0.2035, 0.3259], [0.2407, 0.3805]i

h[0.6207, 0.7490], [0.2049, 0.3275], [0.2410, 0.3819]i

h[0.4953, 0.5967], [0.1726, 0.3167], [0.3001, 0.4017]i

h[0.4940, 0.5960], [0.1735, 0.3175], [0.3005, 0.4022]i

h[0.4951, 0.6277], [0.1923, 0.3249], [0.1641, 0.2870]i

h[0.4899, 0.6247], [0.1948, 0.3290], [0.1647, 0.2882]i

h[0.3641, 0.4658], [0.1325, 0.2432], [0.3647, 0.4521]i

h[0.3621, 0.4641], [0.1327, 0.2437], [0.3672, 0.4547]i

INFWA [20]

INFWG [20]

h[0.3101, 0.4713], [0.1585, 0.2632], [0.3440, 0.4755]i

h[0.2834, 0.4518], [0.1719, 0.2722], [0.3639, 0.4970]i

h[0.6217, 0.7496], [0.2021, 0.3243], [0.2404, 0.3791]i

h[0.6149, 0.7437], [0.2374, 0.3480], [0.2463, 0.3967]i

h[0.4967, 0.5973], [0.1716, 0.3158], [0.2996, 0.4012]i

h[0.4833, 0.5893], [0.2002, 0.3293], [0.3070, 0.4074]i

h[0.5003, 0.6308], [0.1897, 0.3209], [0.1634, 0.2859]i

h[0.4492, 0.5914], [0.2562, 0.3822], [0.1839, 0.3060]i

h[0.3661, 0.4676], [0.1323, 0.2427], [0.3621, 0.4494]i

h[0.3482, 0.4513], [0.1410, 0.2525], [0.3959, 0.4778]i

H-INWAG

H-INOWAG

h[0.2964, 0.4613], [0.1652, 0.2677], [0.3545, 0.4867]i

h[0.2606, 0.4340], [0.1937, 0.2970], [0.3892, 0.5414]i

h[0.6183, 0.7467], [0.2201, 0.3076], [0.2434, 0.3880]i

h[0.6006, 0.7315], [0.2201, 0.3184], [0.2286, 0.3585]i

h[0.4900, 0.5933], [0.1862, 0.3226], [0.3033, 0.4043]i

h[0.5099, 0.6109], [0.2297, 0.3573], [0.3184, 0.4194]i

h[0.4740, 0.6109], [0.2246, 0.3529], [0.1739, 0.2961]i

h[0.4193, 0.5442], [0.3216, 0.4517], [0.2201, 0.3449]i

h[0.3570, 0.4593], [0.1367, 0.2477], [0.3598, 0.4639]i

h[0.3170, 0.4191], [0.1465, 0.2670], [0.3863, 0.4894]i

H.Garg and Nancy

Table 7 Score values of aggregated INNs INWA [28] A1 A2 A3 A4 A5

-0.2898 0.0797 -0.0887 -0.0544 -0.2283 INFWG [20]

A1 A2 A3 A4 A5

-0.2849 0.0009 -0.08565 -0.04385 -0.23385

INWG [28]

INOWA[28]

-0.2582 -0.3328 0.1029 0.2015 -0.0666 0.0094 0.0187 -0.0591 -0.1958 -0.1834 INHWA[17] γ=2 γ=3 -0.3295 -0.3633 0.1245 0.1032 -0.0778 -0.1020 -0.1076 -0.1874 -0.2479 -0.2765

INOWG [28]

INFWA [20]

-0.3963 -0.2299 0.0823 0.1127 -0.1258 -0.0942 -0.2633 0.0856 -0.3046 -0.1764 Proposed operators H-INWAG H-INOWAG -0.2344 -0.2392 0.1099 0.1072 -0.04955 -0.0518 0.0772 0.0689 -0.1807 -0.1860

Table 8 Ordering of the alternatives Existing operators INWA [28] INOWA [28] INWG [28] INOWG [28] INHWA [17] INFWA [20] INFWG [20]

A2 A2 A2 A2 A2 A2 A2

A4 A3 A4 A3 A4 A4 A4

Ordering A3 A5 A4 A5 A3 A5 A4 A5 A3 A5 A3 A5 A3 A5

A1 A1 A1 A1 A1 A1 A1

Proposed operators H-INWAG H-INOWAG

Ordering A2 A4 A3 A5 A1 A2 A3 A4 A5 A1

Table 9 Effect of parameter λ on score values and ranking A1 A2 A3 A4 A5 Ranking

H-SVNWAG -0.2894 0.1516 0.0120 0.0201 -0.1913 (24351)

A1 A2 A3 A4 A5 Ranking

-0.2518 0.1894 0.0464 0.0736 -0.1295 (24351)

Score Values (λ = 0) H-SVNOWAG H-INWAG -0.394 -0.2898 0.1467 0.0797 -0.0438 -0.0887 -0.01556 -0.0544 -0.2549 -0.2283 (24351) (24351) Score Values (λ = 0.7) -0.3511 -0.2453 0.1848 0.1124 -0.0068 -0.0577 -0.0886 0.0488 -0.2053 -0.1826 (24351) (24351)

H-INOWAG -0.3963 0.0823 -0.1258 -0.2633 -0.3046 (23451)

H-SVNWAG -0.2788 0.1623 0.0218 0.0352 -0.1739 (24351)

-0.3499 0.1117 -0.0924 -0.1560 -0.2651 (23451)

-0.2408 0.2004 0.0563 0.0892 -0.1112 (23451)

Score Values (λ = 0.2) H-SVNOWAG H-INWAG -0.3808 -0.2773 0.1574 0.0889 -0.0333 -0.0799 -0.1368 -0.0255 -0.2409 -0.2154 (24351) (24351) Score Values (λ = 0.9) -0.3389 -0.2322 0.1959 0.1218 0.0040 -0.0487 -0.0690 0.0796 -0.1907 -0.1693 (23451) (23451)

H-INOWAG -0.3832 0.0906 -0.1163 -0.2334 -0.2934 (23451)

H-SVNWAG -0.2627 0.1785 0.0350 0.0581 -0.1474 (24351)

-0.3363 0.1202 -0.0827 -0.1239 -0.2537 (23451)

-0.2353 0.2060 0.0613 0.0971 -0.1020 (24351)

Score Values (λ = 0.5) H-SVNOWAG H-INWAG -0.3631 -0.2582 0.1738 0.1029 -0.0174 -0.0666 -0.1081 0.0187 -0.2196 -0.1958 (24351) (24351) Score Values (λ = 1) -0.3328 -0.2257 0.2015 0.1266 0.0094 -0.0442 -0.0591 0.0952 -0.1834 -0.1626 (24351) (24351)

H-INOWAG -0.3633 0.1032 -0.1020 -0.1874 -0.2765 (23451) -0.3295 0.1245 -0.0778 -0.1076 -0.2479 (23451)