1 tayangan

Judul Asli: 06622425

Diunggah oleh Ðânĩĕl JøŚĕ CäŚŧrø Ârïźă

- The Mathematics of Living Systems
- Fuzzy numbers & Fuzzy Equation.pptx
- JSIR 69(1) 11-16
- MET 49-2-99 102 Jurisica Mehatronika1
- Self Generated Fuzzy Membership Function Using
- Fuzzy Logic
- Selection of Equipment by Using Saw and Vikor Methods
- IJETR033040
- 6d61cc12ed3ddc44e69a12baba6fa995d355
- A Course in Fuzzy Systems and Control_part 1
- Paper on Fuzzy Delhi Loads
- Eureka Sample
- Phillips 2011 What is Strategy
- Industrial_Image_Processing_Using_Fuzzy-logic.pdf
- Applying ER-MCDA and BF-TOPSIS to Decide on Eﬀectiveness of Torrent Protection
- Bell Institutions Think
- Principles of Management --- Course Outline
- 00987074
- The Real Challenge of the CEO
- QPC-

Anda di halaman 1dari 8

Syibrah Naim and Hani Hagras

The Computational Intelligence Centre

School of Computer Science and Electronic Engineering

University of Essex, United Kingdom

mnnurs@essex.ac.uk

Abstract Decision making could be viewed to include Multi- Makers (DMs) and we try to aggregate all the individual

Criteria Group Decision Making (MCGDM). MCGDM is a preferences into a collective preference. Hence, it is essential

decision tool which it is able to find a unique agreement from to combine MCDM and GDM to form Multi-Criteria Group

number of decision makers/users by evaluating the uncertain Decision Making (MCGDM). MCGDM plays and important

judgment among them. Several fuzzy logic based approaches role to evaluate the utmost decision among humans

have been employed in MCGDM to handle the linguistic preferences which involves high of uncertainties.

uncertainties and hesitancy. However, there is a need to handle

the high level of uncertainties that exist in decision making According to [2] and [3], there are many sources of

problems involving numbers of decision makers/experts/users uncertainties facing MCGDM systems in real world-

with varying points of view. In this paper, we present a general applications. Such uncertainties include linguistic

type-2 fuzzy logic based approach for MCGDM. The proposed uncertainties where a criterion like Age might be

system aims to handle the high levels of uncertainties which interpreted in different ways according to different DMs. The

exist due to the varying Decision Makers (DMs) judgments hesitancies and vagueness might exist internally and

and the vagueness of the appraisal. The proposed method externally. The internal conflicts such as self-esteem and

utilizes general type-2 fuzzy sets. The aggregation operation in confidence level will affect the DM judgment during the

the proposed method aggregates the various DMs opinions assessment. In addition, external circumstances such as the

which allow handling the disagreements of DMs opinions into political situation, the circumstances prevailing at that time

a unique approval. We will present results from the proposed

and the environmental conditions can definitely have an

system deployment for the assessment of the postgraduate

study. The proposed system was able to model the variation in

effect on DMs opinions.

the group decision making process exhibited by the various Researches have been investigating techniques to handle

decision makers opinions. In addition, the proposed system the faced uncertainties in many decision making

showed agreement between the proposed method and the real applications. Fuzzy logic is regarded as an appropriate

decision outputs from DMs (as quantified by the Pearson methodology for decision making systems which is able to

Correlation) which outperformed the MCGDM systems based simultaneously handle numerical data and linguistic

on type-1 fuzzy sets, interval type-2 fuzzy sets and interval knowledge. Research in fuzzy decision making has grown

type-2 fuzzy sets with hesitation index.

rapidly in the utilization of extended fuzzy set theories (i.e.,

Keywords- general type-2 fuzzy sets, multi-criteria group Intuitionistic Fuzzy Sets (IFSs) [4], Hesitant Fuzzy Sets [5],

decision making Vague Sets [6], Interval-valued Fuzzy Sets [7]). The work in

[8] developed an interactive decision support system for

I. INTRODUCTION sustainable energy management and the application of fuzzy

Uncertainties, vagueness and hesitancy pose major methods to tackle uncertainties in the data. The work

challenges to any decision making system. Uncertainties are presented in [9] studied the supplier selection which involved

not just limited to the linguistic evaluation where several conflicting criteria where the decision makers

uncertainties can be referred to as a situation where the knowledge is usually vague and imprecise.

current state of knowledge is such that the order or nature of The application of Type-2 fuzzy sets in decision making

things is unknown or vague [1]. Furthermore, uncertainties has been widely applied. In [10], a method was proposed to

are present when the consequences, extent, or magnitude of complement the methods presented in [11] and [12] for fuzzy

circumstances, conditions, or events is unpredictable [1]. The multiple attribute group decision-making based on interval

evaluation and the consideration of uncertainties vary type-2 fuzzy sets. In [13], a method was proposed which

according to the given application. could handle evaluating values represented by non-normal

Decision making could be viewed to include Multi- interval type-2 fuzzy sets. The work presented in [14]

Criteria Decision Making (MCDM) and Group Decision investigated group decision making problems in which all

Making (GDM). MCDM aims to provide a selection the information provided by decision makers (DMs) are

involving a number of conflicting alternatives and criteria. expressed as interval type-2 fuzzy values in decision

On the other hand, in GDM, we have a group of Decision matrices.

Recently several researchers began to explore the proposed method aggregates the various DMs opinions

application of general type-2 fuzzy sets and systems. In [15], which allow handling the disagreements of DMs opinions

two methods were proposed for the automatic design of into a unique approval. We will present results from the

general type-2 fuzzy sets using data gathered through a proposed system deployment for the assessment of the

survey on the linguistic variables. A series of results postgraduate study. The proposed system was able to model

presented in [16] related to the different levels of uncertainty the variation in the group decision making process exhibited

handled by the different types of Fuzzy Logic Systems by the various decision makers opinion. In addition, the

(FLSs) including general type-2 fuzzy logic systems. In [17], proposed system showed agreement between the proposed

an approach was presented for uncertain fuzzy clustering method and the real decision outputs from DMs which

using the general type-2 fuzzy C-means algorithm and it was outperformed the MCGDM systems based on type-1 fuzzy

able to balance the performance of type-1 algorithms in sets, interval type-2 fuzzy sets and interval type-2 with

various uncertain pattern recognition tasks. hesitation index.

In our previous work ([18] and [19]), we developed type- The rest of the paper is organized as follows: Section II

1, interval type-2 and interval type-2 fuzzy logic with will present a brief overview on general type-2 fuzzy sets,

hesitant index based on IFSs (Intuitionistic Fuzzy Sets) for type-2 fuzzy logic rule based systems and multi-criteria

MCGDM. Fig. 1, a, b, c, d show respectively the secondary decision making. Section III presents the proposed general

membership functions for the type-1, interval type-2, interval type-2 fuzzy logic based approach for MCGDM (GFL-

type-2 fuzzy sets with hesitant index based on IFSs and MCGDM). Section IV presents an example to clarify the

general type-2 fuzzy sets. It was shown in [19] that the various stages of the proposed method. Section V reports on

system based on interval type-2 fuzzy sets with hesitant the results. Finally, Section VI presents the conclusions and

index based on IFSs could handle the linguistic uncertainties future work.

by the interval type-2 fuzzy set Footprint of Uncertainty

(FOU). In addition this combination simultaneously II. PRELIMINARIES

computes the hesitancy from the membership and non- A. General Type 2 Fuzzy Sets

membership degree (of IFSs). However, the interval values

with hesitation index cannot fully represent the uncertainty A general type-2 fuzzy set (as shown in Fig. 2), denoted

~

distribution (in the third dimension) associated with the A , is characterized by a general type-2 fuzzy membership

decision makers. function A~ ( x, u ) , where x X and J x [0, 1] , i.e.,

~

A = {((x, u), A~ (x, u)) | x X , u J x [0, 1]} (1)

~

in which 0 A~ (x, u ) 1 . A can also be expressed as

follows:

~

A = xX uJ X

A~ (x, u ) (x, u ) J x [0, 1] (2)

where denotes union over all admissible x and u . J x is

~

called primary membership of x in A (as shown in Fig.2a),

(a) (b)

where J x [0, 1] for x X [20]. The uncertainty in the

primary memberships of a general type-2 fuzzy set consists

of a bounded region that is called the Footprint of

Uncertainty (FOU) [20] which is the aggregation of all

primary memberships [17]. According to [9], a general type-

2 fuzzy set can be thought of as a large collection of

embedded type-1 sets each having a weight to associate with

it [21]. At each value of x , say x = x , the 2-D plane whose

axes are u and A~ (x , u ) is called a vertical slice of A~ ( x , u )

(c) (d)

[20]. A secondary membership function is a vertical slice of

A~ ( x , u ) [20]. Hence, A~ (x , u ) for x X and

Figure 1. View of the secondary membership function in the third

dimension, x u plane (a) Type-1 Fuzzy Set. (b) Interval Type-2 Fuzzy Set. u J x [0, 1] could be written as [20]:

(c) Interval Type-2 Fuzzy Set with Hesitation index (from IFSs). (d)

General Type-2 Fuzzy Set. A~ (x = x, u ) A~ (x) = f x (u ) / u (3)

uJ x

In this paper, we present a general type-2 fuzzy logic in which 0 f x (u ) 1 . Because x X , the prime

based approach for MCGDM (GFL-MCGDM) which is

more suited for higher levels of uncertainties. The proposed notation on A~ (x ) is dropped and A~ (x ) is referred to as a

method utilizes general type-2 fuzzy sets to evaluate the secondary membership function [22]; it is a type-1 fuzzy set

linguistic uncertainties within the DMs judgments about the which is also referred to as a secondary set (see Fig. 2b) [23].

linguistic variables. The aggregation operation in the

indifference between xi and x j , x ij > 0 . 5 indicates that xi

is preferred to x j , and x ij < 0 . 5 indicates that x j is

preferred to xi .

MULTI-CRITERIA GROUP DECISION MAKING

(GFL-MCGDM)

The proposed GFL-MCGDM combines general type-2

(a) (b) fuzzy logic systems with MCGDM. In the overview on the

Figure 2. General Type-2 Fuzzy set (a) The primary membership, J x . proposed GFL-MCGDM shown in Fig. 3, the rules and

(b) The secondary membership is a fuzzy set.

fuzzifier are used while the inference engine, the type-

reduction and defuzzification of type-2 FLSs will be replaced

B. Type-2 Fuzzy Logic - Rule Based Systems with the MCGDM method where we rank the output values

from the priority weights.

According to [23], in general type-2 FLSs, the rules will

remain the same as in type-1 FLSs but the antecedents and

the consequents will be represented by general type-2 fuzzy

sets. Consider a type-2 FLS having inputs

,..., and outputs ,..., . The th

rule in this multiple-inputmultiple-output type-2 FLS can be

written as follows: : IF is And . . . And is ,

THEN is . . . is where is the number of rules

in the rule-based.

C. Multi-Criteria Group Decision Making

The MCGDM is explained as follows: Let A be a set of

alternatives, let X be as set of criteria and let D be a set of

experts/DMs, where A = {l1 , l2 ,..., le } , X = {x1 , x2 ,..., xn }

D = {z1 , z2 ,..., z m } , respectively. A MCGDM problem can be

concisely expressed in a matrix format as follows:

x1 x2 xn

x1 x 11 x 12 x1 n

D k = lr = x 2 x 21 x 22 x2n (4)

Figure 3. An overview on the proposed GFL-MCGDM.

In what follows, we will state the basic approach to fuzzy the alternatives represent the FLS outputs. We modified the

MCDM without considering risk attitude and confidence. A reciprocal decision matrices by inserting the fuzzy logic rule-

general decision making problem with e alternatives base. As shown in Fig. 3, the proposed architecture starts by

l r (r = 1,..., e ) , n criteria xt (t = 1,..., n ) and m experts receiving crisp inputs from the criteria which are then

fuzzified and then fire the fuzzy rule-base to provide the

z k (k = 1,..., m ) can be concisely expressed as: D = l r . Here membership values in decision matrices. The membership

D refers to a DM (where the entry xij represents the rating value at a given x for a general type-2 fuzzy set is a type-1

fuzzy set in the third dimension. The aggregation operation

of the rule formed by criteria xi and criteria x j ) where in the proposed method aggregates the various DMs opinions

i = 1,..., n and j = 1,..., n . which allow handling the disagreements of DMs opinions

into a collective approval. The ranking components in the

Definition 1. A preference relation P on the set X is proposed GFL-MCGDM utilize fuzzy arithmetic averaging

characterized by a function P : X X U , where U is the operators to compute the membership values in decision

domain representation of preference degrees. A fuzzy matrices. The normalizations and the priority weights will be

preference relation P on the set X is represented by a calculated to determine the final output. The various

complementary/reciprocal matrix : components of the proposed GFL-MCGDM will be

discussed in the following subsections.

( )

X = xij nn

X X with x ij 0 , x ij + x ji = 1, x ii = 0 . 5 for all

A. Generation of the General Type-2 Fuzzy sets for the

i, j = 1, 2,...,n where x ij denotes the preferred degree of the Lingusitic Labels for each Criterion

criteria xi over xj . In particular, xij = 0.5 indicates All fuzzy sets representing the linguistic labels for each

criterion were modelled in the x-u domain with trapezoidal

type-2 fuzzy membership functions (as shown in Fig. 4a). B. The GFL-MCGDM Steps

Essentially, the linguistic labels type-2 fuzzy sets are created This section presents eight steps to determine the ranking

by the combination of DMs opinions (modelled by of the outputs. This phase included the fuzzifier process and

symmetrical triangular type-1 fuzzy sets as shown in Fig. decision-making process. The steps below provide an

4a). The minimum, maximum and the average values defined overview of the steps of the proposed GFL-MCGDM:

by the intersection of DMs opinion are demarcated to create

Step 1: Consider a multi-criteria group decision making

the support for each trapezoidal type-2 fuzzy set.

problem, let A = {l1 , l2 ,..., le } be a discrete set of alternatives

(output parameters), X = {x1 , x 2 ,..., x n } be a set of criteria

(input parameters), and D = {z1 , z 2 ,..., z m } be a set of DMs.

The DM z k D provides his/her judgment based on the

rules given, and constructs the rule-based reciprocal decision

matrix.

Step 2: With the assumption that we have the input

values for each criteria, we utilize general type-2

membership function (explained in the previous subsection

and shown in Fig. 4) to define the membership degree for

(a) (b) each rule defined in the reciprocal decision matrices and we

Figure 4. Generation of General Type-2 Fuzzy Set from DMs opinion (a) then identify the rule that is fired.

Primary membership function (plotted in thick lines) as generated from the

DMs type-1 Fuzzy Sets (plotted in thin dashed lines) and the used type-1 Each x ij (r , k ) calculated in the decision matrix should

Fuzzy Sets for comparison in thick dashed line. (b) A secondary satisfy the following rule according to DMs opinions:

membership functions in the third dimension.

~ ~

IF xi is Fi and x j is F j THEN l r (10)

Each DM opinion about a given linguistic label is defined

by a symmetrical type-1 triangular fuzzy set (as shown in (

X (r , k ) = x ij (r , k ) ) (11)

Fig. 4a) which is defined by three points (a, b, c). We nn

aggregate a = {a 1 , ... , a m } , b = {b1 , ... , b m } and Hence, for each x ij ( r , k ) , we will have ij(r , k ) , (r , k ) , for all

ij

c = {c 1 , ... , c m } to find the lowest, vertex and upper points i , j = 1, 2,..., n .

of the generated type-2 fuzzy set according to the number of

decision makers (k = 1,..., m ) . To draw the x-u domain of the ~(r , k ) as follows:

Step 3: In this step, we define ij

variable, the following points have to be defined as follows: ij(r , k ) + ij(r , k )

~ij(r , k ) = (12)

2

a = min {a 1 , ... , a m } (5)

Hence, for each entry, we will have

b = min {b1 , ... , b m }

b ' ' = max {b1 , ... , b m }

(6)

(7)

(r , k )

(

(r , k ) ~ ( r , k ) ( r , k )

xij = ij , ij , ij )

, for all i , j = 1,2,..., n .

c = max {c1 , ... , c m } (8) In all the operations below please note that all operations

b + b on x will be carried on , ~ and independently in

d' = (9)

2 decision matrices.

Step 4: Then, we use the min operator to compute the

Thus, the above equations result in generating the

firing strength for each rule. This will lead to construct the

general type-2 fuzzy set for each label from the individual

fuzzy decision matrices. Based on the DMs/experts z k D ,

DM opinions (represented as type-1 fuzzy sets). The

secondary membership function at each x is a symmetrical we can construct reciprocal decision matrices.

triangle (as shown in Fig. 4b). Step 5: The general type-2 fuzzy values of each x ij

(r , k )

Thus, the generated general type-2 fuzzy set (in Fig.4a) are then aggregated. The aggregated set can be determined

upper membership function will be formed by points a, b, by x ij(r ) = (v ijr , w ijr , y ijr ) for (k = 1,..., m ) where,

b and c. While the lower membership function will be

formed by points b, d and b. The type-1 fuzzy sets v ijr = min

k

{ ij

(r , k )

} (13)

(which will be used when comparing the performance of a

type-1 fuzzy based system with the proposed system) will 1 m ~ (r , k )

w ijr = ij (14)

consist of the points e (average of a and b), f(average of m k =1

b and d), g (average of b and d) and h (average of b

and c) as shown in Fig. 4a. y ijr = max

k

{ ( )} ij

r ,k

(15)

Step 6: Use the fuzzy arithmetic averaging operator to TABLE I. EXAMPLE OF A RULE FOR ALTERNATIVE FOREIGN FROM

aggregate all x ij( r ) = (v ijr , w ijr , y ijr ) corresponding to the n A DECISION MAKER

Financial (F) Very Expensive (VE)

1 n (r )

xt(r ) = xij (16) Age (A) Young (Y)

n t =1 Distance according to flight hours (D) Medium Distance (MD)

Step 7: Find the average of each xt( r ) (where t = 1,..., n ), Number of Dependent (N) Many (A)

(r )

this average is called xtavg . Next, normalize the matrix so

that each element in the matrix can be written as follows: Table I shows a rule for the Foreign alternative which

is according to DM 1: IF the Financial is Very Expensive,

(r )

(r )

xtavg Age is Young, Distance is Medium and Number of

xtnorm = (r ) (17) Dependent is Many THEN further Foreign Study.

tn=1 xtavg

A. Analysis according to the Proposed Steps

Step 8: Find the priority weights, l r of each alternative This section presents the eight steps involved in the

as: proposed GFL-MCGDM to determine the ranking of the

n outputs. All of the steps are simultaneously involved in the

lr = 1 x( ) r

tnorm

(18) decision making process in order to determine the fuzzy

n t =1 ranking. The generated general type-2 fuzzy sets are shown

r in Fig.5.

where l r > 0 , r = 1, , e , er =1 l = 1 .

IV. AN EXAMPLE TO CLARIFY THE PROPOSED Step 1: According to the first step, we construct the

GFL-MCGDM reciprocal decision matrices based on the fuzzy rules set

collected from all decision makers. According to the ten

In order to evaluate the proposed GFL-MCGDM system, DMs evaluated, their opinions were based on the three output

we have conducted a survey among postgraduate students in variables/alternatives (Foreign, Local and Internship).

the Department of Computer Science, University of Essex. Consequently, we constructed 30 matrices because each of

Ten participants were chosen randomly and they were

the rules represents the experts/DMs opinion. 30 matrices

classified as decision makers (DMs) for the group decision-

will be accumulated in this analysis to compute one decision.

making system. In this analysis, DMs are chosen based on

their knowledge and experience in choosing a university for Hence, the multi-criteria group decision making problem

their postgraduate study. The group of DMs consists of local can be formulated as having A = {l1 , l2 , l3} to be a discrete

students, foreign students and internship students. The set of alternatives (output parameters), X = {x1, x2 ,. x3, x4} be

system will determine three types of decisions which are

foreign postgraduate study, local postgraduate study and a set of criteria (input parameters), and D = {z1 , z2 ,...,z10} be

internship postgraduate study. Based on the DMs judgment a set of DMs. The DM zk D provides a judgment based on

and assessment, we generate the system.

the rules given and constructs the decision matrix.

Three output fuzzy variables were used in this system:

study Foreign, study Local and study Internship. The Using the associated abbreviations in Table I, for

developed decision system provides a decision output alternative Foreign ( r = 1 ) and for DM1( k = 1 ), we

(foreign, local or internship) based on the input criteria from construct a rule-based reciprocal decision matrix as follows:

the students. The fuzzy system for postgraduate decisions criteria F A D N

was based on four main input variables (criteria); Financial F VE and Y VE and MD VE and A

support (cost of living per month that they need to consider); A VE and Y Y and MD Y and A

Foreign =

Age; preferred living Distance from hometown to the D VE and MD Y and MD MD and A

university (flight hours); Dependents (how many dependents N VE and A Y and A MD and A

they need to support during their postgraduate study).

The linguistic labels for Financial are Very Cheap, For example, for DM1, we can construct the rule set as

Cheap, Medium, Expensive, Very Expensive. The linguistic follows: Let us say that criterion xi =1 is Financial (F),

labels for Age are Very Young, Young, Medium, Old, Very criterion xj=3 is Distance (D) and alternative is study

Old. The linguistic labels for Distance according to flight

1,1

hours are Very Near, Near, Medium, Far, Very Far. The Foreign. In this case, the rule set for x13 (shown in bold in

linguistic labels for the Number of Dependents are Very the table above) is denoted as IF your Financial (F) is Very

Little, Little, Medium, Many, So Many. Expensive (VE) and Distance (D) is Medium (MD), THEN

the you should further Foreign study.

Step 2, Step 3 and Step 4: In case the input values are: Fired Rule x 13(r = 1 ) = (v 13r = 1 , w 13r = 1 , y 13r = 1 )

financial support is 1000, age is 26 years old, preferred

Financial (VE) vs. Distance (MD) [0, 0.1919, 1.0]

distance according to flight hours is 4 hours and number of

dependents is 0. From Fig. 5 (showing only the generated

type-2 fuzzy sets for Financial and Distance) and the above Whereas for elements such as x121 , x131 , x32

1

, x141 , x142 , x143 ,

1,1

matrix, general type-2 fuzzy value for x13 could be written as the reciprocal or complementary value are utilized. For

~

follows (Equation (12) is used to define ): example, the complimentary vertex elements for x131 is x311 .

1

Membership ~ Thus, the complimentary vertex value for w31 is 1 0.1919

General type-2 fuzzy values ( , and )

Values = 0.8081. This is because we are using the fuzzy

Financial (VE) (0, 0.2, 0.4) complementary matrix (refer to Definition 1). Whereas for

Distance (MD) (0.6667, 0.8333, 1.0 )

r

the value xii will be 0.5 as according to Definition 1.

r =1

The calculations for the remaining elements xij will

follow the same calculations above. Hence,

x ij(r = 1 ) = (v ijr = 1 , w ijr = 1 , y ijr = 1 ) for the remaining elements by

all ten DMs to represent alternative Foreign (r = 1) could be

found as follows:

vij1

x12 0 x 21 1.0

x 23 0 x 41 1.0

x 24 0 x 42 1.0

Figure 5. Example of general type-2 fuzzy membership function for

criteria financial (very expensive, for x=1000 ) and distance (medium, for x34 0 x 43 1.0

x=4 hours).

1

By using the min operator, the firing strength for the rule

If Financial (VE) and Distance (MD) Then Foreign in 0.2493 0.7507

x12 x 21

the reciprocal decision matrix (from the above matrix) is:

x13 0.1919 x31 0.8081

Fired Rule ~

Firing Strength ( , and )

x14 0 x32 0.8104

Financial (VE) and (0, 0.2, 0.4)

Distance (MD) x 23 0.1896 x 41 1.0

x 24 0 x 42 1.0

Step 5: The general type-2 fuzzy values for x131,k over the 10

x34 0 x 43 1.0

DMs ( k = 1,..., 10 ) for alternative Foreign are as follows:

Decision Firing Strength for Financial (VE) versus Distance

Maker

(MD),

1, k

x13 ~

( , and )

y ij1

x12 1.0 x 21 0

1 (0, 0.2, 0.4)

2 (0.5714, 0.7857, 1.0) 1.0 0

3 (0, 0.2, 0.4) x13 x31

4 (0, 0.1667, 0.333) 0 0

5 (0, 0, 0) x14 x32

6 (0, 0, 0) 1.0 1.0

7 (0, 0.2, 0.4) x 23 x 41

8 (0, 0.2, 0.4) 0 1.0

9 (0, 0.1667, 0.333)

x 24 x 42

10 (0, 0, 0) x34 0 x 43 1.0

Thus, according to Equations (13), (14) and (15), The matrices for alternatives Local ( r = 2 ) and

= (v 13r = 1 , w 13r = 1 , y 13r = 1 ) will be generated for x13 for

(r = 1 ) 1

x 13 Internship ( r = 3 ) could be easily found by following the

alternative Foreign by ten DMs, as follows: calculations employed for the alternative Foreign.

Step 6: Next, we average the above cumulative matrices Step 8: In the last step, we compute the priority weights. For

for x ij(r ) = (v ijr , w ijr , y ijr ) . The following example will show example for Foreign alternative using Equation (18):

only for the lower membership values, vij1 . Hence, for 0 .3751 + 0 .3244 + 0 .3139 + 0 .3333

l1 = = 0 . 3367

4

criterion Financial (t = 1) alternative Foreign (r = 1) , for

vij1 , using Equation (16), we have: Thus: l 1 = 0.3367 , l 2 = 0.3315 and l 3 = 0.3318

1

v 11 1

+ v 12 1

+ v 13 1

+ v 14 0 .5 + 0 + 0 + 0 The weighted values allow us to rank the alternatives and

v 11 = = = 0 . 125 the highest ranking can be determined as an output/decision.

4 4

Thus, alternative l 1 has the highest value. The GFL-

Thus, the average cumulative matrices for x t(=r (=11,..., 2 4, 3)) for each

MCGDM suggests that IF the input values of financial

v , w , y , respectively is written as follows: support is 1000, age is 26 years old, preferred distance

v w y according to flight hours is 4 hours, and number of

dependent is 0 THEN then you should further your study in

t = 1 0.1250 , 0 . 2352 , 0 . 6250

Foreign Foreign.

t = 2 0 . 3750 , 0 . 3601 , 0 . 3750

t = 3 0 . 6250 , 0 . 5296 , 0 . 1250 V. EXPERIMENTS AND RESULTS

t = 4 0 . 8750 , 0 . 8750 , 0 . 8750 The efficiency of the proposed system can be evaluated

v w y through the correlation values between the DMs decision

t = 1 0 . 1250 , 0 . 1584 , 0 . 5312 and the output ranking. In this work, Pearson Correlation

Local

t = 2 0 . 3750 , 0 . 4132 , 0 . 3750

was used to find the correlation between the DMs decision

t = 3 0 . 6250 , 0 . 5533 , 0 . 2187 and the various MCGDMs decision is as follows:

t = 4 0 . 8750 , 0 . 8750 , 0 . 8750 COV ( X , Y ) E [( X X )(Y Y )] (19)

X ,Y = =

v w y X Y X Y

t = 1 0 . 1250 , 0 . 1704 , 0 . 5313

Internship

t = 2 0 . 3750 , 0 . 3989 , 0 . 3750

The aforementioned values show that the higher the

t = 3 0 . 6250 , 0 . 5557 , 0 . 1250 value, the closer the proposed system to the DMs decision.

t = 4 0 . 8750 , 0 . 8750 , 0 . 8750 We tested 30 data sets to find the correlation values between

the linguistic decision from the DMs and the output from the

Step 7: Then, we find the average value of each xt( r ) , hence proposed GFL-MCGDM system. Examples of three data sets

for example for x11 , the average could be found as with the input values for each criterion (Financial (F), Age

(A), Distance (D) and Dependent (N)) and the real output

follows: x11(avg ) = 0 .125 + 0 .2352 + 0 .625 = 0 .3284 decision from the DMs and the proposed systems are shown

3

in Table II. The ranking from both sides will determine the

Following the same approach, we can find the other values, agreement among one another.

hence:

t = 1 0.3284 t = 1 0.2716 t = 1 0.2756 TABLE II. EXAMPLE OF INPUT FOR EACH CRITERION, OUTPUT FROM

Foreign t = 2 0.3700 Local t = 2 0.3877 Internship t = 2 0.3830 DECISION MAKERS AND OUTPUT FROM GFL- MCGDM SYSTEM

t = 3 0.4265 t = 3 0.4657 t = 3 0.4665 F A D N Output Output

() Age (hours) Dependent Decision GFL-

t = 4 0.8750 t = 4 0.8750 t = 4 0.8750

Makers MCGDM

1000 26 4 0 Foreign Foreign

We then find the normalized values following Equation (17). 1100 22 0.5 0 Internship Foreign

For example for the alternative Financial:

800 22 0.75 0 Local Local

1 0 .3284

x1( norm ) = = 0 .3751

0 .3284 + 0 .2716 + 0 .2756 According to Table III, it can be observed that type-1

The normalization gives the following matrices: fuzzy based MCGDM gives 0.1299 correlations to the

linguistic appraisal of the DMs (i.e. the DMs decision)

r = 1 0.3751 r = 1 0 . 3244 whereas interval type-2 fuzzy based MCGDM gives a

Financial r = 2 0 . 3101 Age r = 2 0 . 3399 correlation value of 0.3928. In addition, the interval type-2

r = 3 0 . 3147 r = 3 0 . 3357 fuzzy logic with hesitation index based MCGDM gives a

correlation value 0.4787. Markedly, the proposed GFL-

r = 1 0 . 3139 r = 1 0 . 3333 MCGDM gives the highest correlation value of 0.5148.

Distance r = 2 0 . 3427 Dependent r = 2 0 . 3333 Hence, the proposed system was able to model the variation

r = 3 0 . 3433 r = 3 0 . 3333 in the group decision making process exhibited by the

various decision makers opinion. In addition, the proposed

system showed agreement between the proposed method and

the real decision outputs from DMs (as quantified by the [5] R. M. Rodrguez, L. Martnez, and F. Herrera, Hesitant Fuzzy

Linguistic Term Sets for Decision Making, IEEE Transactions

Pearson Correlation) which outperformed the MCGDM On Fuzzy Systems, vol. 20, no. 1, pp. 109-119, 2012.

systems based on type-1 fuzzy sets, interval type-2 fuzzy sets [6] R. R. Yager, Fuzzy decision making including unequal

and interval type-2 with hesitation index. objectives, Fuzzy Sets and Systems, vol.1, pp.87-95, 1978.

[7] L. A. Zadeh. Fuzzy sets, Information and Control, vol. 8, pp.

TABLE III. PEARSON CORRELATION VALUES FOR DIFFERENT TYPE 338353, 1965.

OF FUZZY SETS [8] S. D. Pohekar, Application of multi-criteria decision making to

sustainable energy planningA review, Renewable and

Methods Pearson Correlation

Sustainable Energy Reviews, vol. 8, no. 4, pp. 365381, 2004.

Type 1 fuzzy Sets 0.1299

[9] F. E.Borana, S. Gena, M. Kurtb, and M. Kurtb, A multi-criteria

Interval Type 2 Fuzzy Sets 0.3928 intuitionistic fuzzy group decision making for supplier selection

Type 2 Fuzzy Sets with Hesitation Index 0.4787 with TOPSIS method, Expert Systems with Applications, vol.36,

General Type 2 Fuzzy Sets 0.5148 no. 8, pp. 1136311368, 2009.

[10] S. M. Chen, M. W. Yang, L. W. Lee, and S. W. Yang, Fuzzy

VI. CONCLUSIONS AND FUTURE WORK multiple attributes group decision-making based on ranking

interval type-2 fuzzy sets, Expert Systems with Applications,

In this paper, we presented a general type-2 fuzzy logic vol. 39, pp. 52955308, 2012.

based approach for MCGDM. The proposed system aims to [11] S. M. Chen, and L. W. Lee, Fuzzy multiple attributes group

handle the high levels of uncertainties which exist due to the decision-making based on ranking values and the arithmetic

varying Decision Makers (DMs) judgments and the operations of interval type-2 fuzzy sets, Expert Systems with

Applications, vol. 37, no. 1, pp. 824833, 2010.

vagueness of the appraisal. The proposed method utilizes

[12] S. M. Chen, and L. W. Lee, Fuzzy multiple attributes group

general type-2 fuzzy sets. GFL-MCGDM provides simple decision-making based on interval type-2 TOPSIS method,

steps which aggregate a group of DMs disagreements Expert Systems with Applications, vol. 37, no. 4, pp. 27902798,

comprehensively into a unique approval. 2010

[13] S. M. Chen, and L. W. Lee, Fuzzy multiple criteria hierarchical

We have presented results from the proposed system group decision-making based on interval type-2 fuzzy sets,

deployment for the assessment of the postgraduate study IEEE Transactions On Systems, Man, And CyberneticsPart A:

where the evaluation involved 10 candidates. The proposed Systems And Humans, vol. 40, no. 5, pp. 1120-1128, 2010.

system was able to model the variation in the group decision [14] W. Wang, X. Liu, and Y. Qin, Multi-attribute group decision

making process exhibited by the various decision makers making models under interval type-2 fuzzy environment,

opinions. In addition, the proposed system showed Knowledge Based Systems, vol. 30, pp. 121-128, 2012.

agreement between the proposed method and the real [15] C. Wagner, and H. Hagras, Novel Methods for the Design of

General Type-2 fuzzy Sets based on Device Characteristics and

decision outputs from DMs (as quantified by the Pearson Linguistic Labels Surveys, IFSA-EUSFLAT , ISBN: 978-989-

Correlation) which outperformed the MCGDM systems 95079-6-8 pp 537-543, pp. 537-543, 2009.

based on type-1 fuzzy sets, interval type-2 fuzzy sets and [16] C. Wagner, and H. Hagras, Toward General Type-2 Fuzzy

interval type-2 with hesitation index. Logic Systems Based on zSlices, IEEE Transactions On Fuzzy

Systems, vol. 18, no. 4, pp. 637-660, 2010.

The increased correlation value shows that the proposed [17] O. Linda, and M. Manic, General Type-2 Fuzzy C-Means

method is considered to be effective in handling the high Algorithm for Uncertain Fuzzy Clustering, IEEE Transactions

level of uncertainties among the DMs. Hence, this shows that On Fuzzy Systems, vol. 20, no. 5, pp. 883-897, 2012.

the proposed method can play an important role in the [18] S. Naim, and H. Hagras, A Fuzzy Logic Based Multi-Criteria

production of better MCGDM which is able to better settle Group Decision Making System for the Assessment of Umbilical

conflicts among the different individual preferences with Cord Acid-Base Balance, WCCI 2012 IEEE World Congress on

Computational Intelligence, Brisbane, Australia, June, 10-15,

different alternatives and criteria followed by synthesizing 2012, pp. 2122 2129.

the different individual preferences into a unanimous

[19] S. Naim and H. Hagras, A Hybrid Approach for Multi-Criteria

approval. Group Decision Making Based on Interval Type-2 Fuzzy Logic

and Intuitionistic Fuzzy Evaluation, WCCI 2012 IEEE World

For future work, we intend to evaluate the proposed Congress on Computational Intelligence, Brisbane, Australia,

method in different real world applications in order to June, 10-15, 2012, pp. 1066 1073.

investigate the efficiency of the proposed method. [20] J. M. Mendel, and R. I. B. John, Type-2 fuzzy sets made

simple, IEEE Transactions On Fuzzy System, 10, 2, 117-127,

VII. REFERENCES 2002.

[1] BusinessDictionary.com (2013, January 14) [online]. Available: [21] Q. Liang, N. N. Karnik, and J. M. Mendel, Connection

http://www.businessdictionary.com/definition/uncertainty.html. admission control in ATM networks using survey-based type-2

[2] S. Naim, and H. Hagras, Type-2 Fuzzy Logic in Multi-criteria fuzzy logic systems, Systems, Man, and Cybernetics, Part C:

Group Decision Making with Intuitionistic Evaluation, Applications and Reviews, IEEE Transactions, 30, 3, 329-339,

Proceeding of the 11th Annual Workshop on Computational 2000.

Intelligence, at the University of Manchester, on 7th 9th [22] H. Hagras, A hierarchical Type-2 Fuzzy Logic Control

September 2011, pp. 1-6. Architecture for Autonomous Mobile Robots, IEEE

[3] F. Herrera, and E. Herrera-Viedma, Linguistic decision analysis: Transactions on Fuzzy Systems, 12, 4, 524-539, 2004.

steps for solving decision problems under linguistic information, [23] J. M. Mendel, Uncertain Rule-Based Fuzzy Logic Systems:

Original Research Article Fuzzy Sets and Systems, vol. 115, no. Introduction and New Directions, Upper Saddle River, NJ:

1, pp. 67-82, 2000. Prentice-Hall, 2001.

[4] K. Atanassov, Intuitionistic fuzzy sets, Fuzzy Sets and Systems,

vol. 110, pp. 87 96, 1986.

- The Mathematics of Living SystemsDiunggah olehmohamedriaz437985
- Fuzzy numbers & Fuzzy Equation.pptxDiunggah olehvpkvikas
- JSIR 69(1) 11-16Diunggah olehathangarasu
- MET 49-2-99 102 Jurisica Mehatronika1Diunggah olehMilos Bubanja
- Self Generated Fuzzy Membership Function UsingDiunggah olehRimbun Ferianto Sr
- Fuzzy LogicDiunggah olehCybergeek Varun
- Selection of Equipment by Using Saw and Vikor MethodsDiunggah olehAnonymous 7VPPkWS8O
- IJETR033040Diunggah oleherpublication
- 6d61cc12ed3ddc44e69a12baba6fa995d355Diunggah olehHassaan Haider
- A Course in Fuzzy Systems and Control_part 1Diunggah olehAstrie Mamanya Alif
- Paper on Fuzzy Delhi LoadsDiunggah olehBhupinder Singh
- Eureka SampleDiunggah olehAnkur Gupta
- Phillips 2011 What is StrategyDiunggah olehJuan Sebastian Quintero Botero
- Industrial_Image_Processing_Using_Fuzzy-logic.pdfDiunggah olehtweenturbo
- Applying ER-MCDA and BF-TOPSIS to Decide on Eﬀectiveness of Torrent ProtectionDiunggah olehMia Amalia
- Bell Institutions ThinkDiunggah olehDavid Lagunas Arias
- Principles of Management --- Course OutlineDiunggah olehMuhammad Salman Rasheed
- 00987074Diunggah olehGanesh Kumar Arumugam
- The Real Challenge of the CEODiunggah olehTulio Takemae
- QPC-Diunggah olehHisoka
- Fuzzy SupervisorDiunggah olehAlejandroHerreraGurideChile
- Fault Diagnosis of an Industrial Gas Turbine Using Neuro-fuzzy MethodsDiunggah olehchristianfv5
- Cluster is a Group of Objects That Belongs to the Same ClassDiunggah olehkalpana
- Quasi TautologyDiunggah olehRaja Sekaran S
- Chapter 11.pptDiunggah olehMoahmed sheikh abdullahi
- OPR 517 Decision Making SkillsDiunggah olehLilian Iordache
- Efficient Fuzzy Set Theoretic Approach to Image Corner MatchingDiunggah olehEditor IJRITCC
- 2012 Lecture 3 Chapter 7 JV Vers 1Diunggah olehKimmi Gounden
- Decision Quality. the Fundamentals of Making Good DecisionsDiunggah olehDorin Miclaus
- Vektor InduksiDiunggah olehajhannany

- FórmulasDiunggah olehÐânĩĕl JøŚĕ CäŚŧrø Ârïźă
- Horarios Formulación y TransferDiunggah olehÐânĩĕl JøŚĕ CäŚŧrø Ârïźă
- Test ExcelDiunggah olehAnonymous sG3iVImJ
- HISTORY.TXTDiunggah olehbobb
- Tayrona2Diunggah olehÐânĩĕl JøŚĕ CäŚŧrø Ârïźă
- composicioncorporal-160909190819Diunggah olehAndreita EC
- Promedio Acumulado.xlsxDiunggah olehÐânĩĕl JøŚĕ CäŚŧrø Ârïźă
- Archivo de Excel y ExposicionesDiunggah olehÐânĩĕl JøŚĕ CäŚŧrø Ârïźă
- reporteHorarioPensum.pdf.pdfDiunggah olehÐânĩĕl JøŚĕ CäŚŧrø Ârïźă
- TayronaDiunggah olehÐânĩĕl JøŚĕ CäŚŧrø Ârïźă
- logoDiunggah olehÐânĩĕl JøŚĕ CäŚŧrø Ârïźă
- Acta De Acuerdos 13-ene.-2019 14-51-50.pdfDiunggah olehÐânĩĕl JøŚĕ CäŚŧrø Ârïźă
- Estudiar para un examen un día antesDiunggah olehÐânĩĕl JøŚĕ CäŚŧrø Ârïźă
- Long, Hei - 21 Techniques of Silent KillingDiunggah olehÐânĩĕl JøŚĕ CäŚŧrø Ârïźă
- Prestamo de PlazoletaDiunggah olehÐânĩĕl JøŚĕ CäŚŧrø Ârïźă
- Talleres de Formacion Integral Def. 2018-1Diunggah olehÐânĩĕl JøŚĕ CäŚŧrø Ârïźă
- 37 Energy Conservation and Renewable Technologies for Buildings to f 2017 SolarDiunggah olehÐânĩĕl JøŚĕ CäŚŧrø Ârïźă
- Malla Curricular Ingeniería Química UniatlánticoDiunggah olehÐânĩĕl JøŚĕ CäŚŧrø Ârïźă
- Hacking The XboxDiunggah olehPanagiotis Tabakis
- 32 Assessment of Bamboo Application in Building Envelo 2017 Construction and BuDiunggah olehÐânĩĕl JøŚĕ CäŚŧrø Ârïźă
- perfil sicviDiunggah olehÐânĩĕl JøŚĕ CäŚŧrø Ârïźă
- BD Estudiantes In Para EncuestaDiunggah olehÐânĩĕl JøŚĕ CäŚŧrø Ârïźă
- Convocatoria Pilo PagaDiunggah olehÐânĩĕl JøŚĕ CäŚŧrø Ârïźă
- 36 Multi Objective Optimization of Building Envelope Design Fo 2016 Energy andDiunggah olehÐânĩĕl JøŚĕ CäŚŧrø Ârïźă
- Cartilla Jurados Nacional Elecciones Congreso 2018Diunggah olehÐânĩĕl JøŚĕ CäŚŧrø Ârïźă
- 38 Commercial Building Energy Use in Six Cities in Southern C 2013 Energy Polic (1)Diunggah olehÐânĩĕl JøŚĕ CäŚŧrø Ârïźă
- moto zDiunggah olehÐânĩĕl JøŚĕ CäŚŧrø Ârïźă
- El Ser Es Uno y Uno El SaberDiunggah olehÐânĩĕl JøŚĕ CäŚŧrø Ârïźă
- Actividad 2Diunggah olehÐânĩĕl JøŚĕ CäŚŧrø Ârïźă
- GUÍA DE LABORATORIO N°2TÉCNICAS GENERALES DE RECOLECCIÓN Y CONSERVACIÓN DE INVERTEBRADOSDiunggah olehÐânĩĕl JøŚĕ CäŚŧrø Ârïźă

- Primates and ParasitesDiunggah olehVictor Andres Garcia
- Oracle v. Santa Cruz Planning Dept MFRDiunggah olehNorthern District of California Blog
- Kinetics of Growth and Sugar Consumption in YeatsDiunggah olehSimone Bassan Zuicker Elizeu
- Manajemen perubahanDiunggah olehIan Pratama
- Security UpgradeDiunggah olehSuneetha Vejandla
- Antiseptic Cream 10 Full TextDiunggah olehanon-46991
- 45255878-Plant-Layout-Final-AKRSVU.txtDiunggah olehMuhammad Naveed
- 47 Top StrategiesDiunggah olehAviel Mizrachi
- SAT April 2018Diunggah olehRockey
- ambush marketingDiunggah olehapi-384489541
- Honey Spot_Teachers Resources.pdf.pdfDiunggah olehJimmy Mehta
- MethergineDiunggah olehJohn Alano Lumberio
- Money Laundering Prohibition) ActDiunggah olehNeosnibor
- Liberal vs. ComelecDiunggah olehCon Pu
- Stress Unmasked, by David Brown extractDiunggah olehSlovenian Webclassroom Topic Resources
- Obama Health Care Plan - Lyndon LaRoucheDiunggah olehpaulpaul
- 2012 Qlogic Storage Area Networking Interoperability GuideDiunggah olehRaul Rodriguez
- LU500 Tutorial Script FinalDiunggah olehjgquevedo
- Atlas Shrugged in AmericaDiunggah olehElliot Young
- All the Cheats - Sims 3Diunggah olehMIKI_37
- Faculty of Business Management(1)Diunggah olehElle J. Johnny
- Xtinction.pdfDiunggah olehArnav Gupta
- antigone ap litDiunggah olehapi-407743290
- Chapter 4 Karma AbkbsDiunggah olehthientrac
- Working Memory - A Guide for SENCOs and Teachers.dec2015Diunggah olehravibhargavaraam
- Chakra Meditation With Mudra and MantraDiunggah olehm_franz
- Frank Lloyd Wright on the Soviet UnionDiunggah olehJoey Smithey
- Barker 2007 - The Science of Harmonics in Classical GreeceDiunggah olehFederico Pantufla
- On Performatives in Legal DiscourseDiunggah olehmha2000
- Medford 0424Diunggah olehelauwit