RETAIL LOCATION DECISION USING AN

INTEGRATED DEMATEL-ANP METHOD

By
ROHAN DESAI
ID no. – 2010B2A4768P

Prepared in partial fulfillment of

Thesis
(BITS C422)

Under the supervision of

Dr. SATYENDRA SHARMA
Department of Management
BITS-Pilani

BIRLA INSTITUTE OF TECHNOLOGY AND SCIENCE

PILANI, INDIA
(Second semester 2014-2015)

1
ACKNOWLEDGEMENT
I am very much thankful to the Dean, Instruction division for giving me this opportunity to do
the project. This work would not be possible without the knowledgeable guidance and immense
support extended by Dr. Satyendra Sharma. I would also like to take this opportunity to thank
Dr. Anil Bhat, Head of Department, Department of Management, for approving this project. A
special thanks to Instructor-in-Charge for Project-type Courses and the administration of BITS
Pilani for letting me do this knowledge enhancing project. I am indebted to my friends and
family for being my source of motivation and having faith in me.

2
CERTIFICATE

This is to certify that the Thesis entitled “Retail location decision using an integrated
DEMATEL-ANP method” and submitted by Rohan Desai bearing ID no. 2010B2A4768P in
partial fulfillment of the requirement of BITS C421T/422T Thesis embodies the work done by
him under my supervision.

Signature of the Supervisor

Dr. Satyendra Sharma
Assistant Professor

3
THESIS ABSTRACT

Purpose: This study focuses on developing a model for making retail location decisions which
addresses the limitations of existing methods. The integrated DEMATEL-ANP method used here
successfully accounts for interdependencies between the selected criteria.

Methodology: After reviewing the existing methods for decision making, the Decision Making
Trial and Evaluation (DEMATEL) and the Analytical Network Process (ANP) methods were
chosen. DEMATEL evaluates the degrees of influence the criteria have on each other. ANP is
used to assign weights to criteria and the sub-criteria, and eventually derive the weights for the
alternative locations.

Findings/ practical implications: The integrated model helps prioritize among the available
alternatives based on scientific methods, much more reliable than intuition and experience based
methods still practiced today. Adopting this method can help retail chains make more informed
decisions.

Limitations: Accuracy of the assumptions made about the alternatives is poor due to the lack of
field work. The accuracy can be vastly improved by consulting with experts and conducting field
research.

Originality: This is the first time an integrated DEMATEL-ANP method has been applied in the
context of retail location decision.

4
TABLE OF CONTENTS

1. INTRODUCTION......................................................................................................................6

2. LITERATURE REVIEW .........................................................................................................7

2.1.EXISTING METHODS AND STRATEGIES.....................................................................7
2.2.APPLICATION OF DEMATEL AND ANP METHODS.................................................12
2.3.CRITERIA USED FOR RETAIL LOCATION DECISION..............................................13

3. DEMATEL METHOD ............................................................................................................14

4. ANP METHOD ........................................................................................................................17

5. RESULT AND CONCLUSION ..............................................................................................25

6. REFERENCES .........................................................................................................................26

5
1. INTRODUCTION

The location of a retail store plays a pivotal role in its success. It is an integral and a crucial part
of the retail strategy because it directly influences the merchandising mix and the store layout.
Once a store is opened, a retailer can take strategic steps towards adjusting prices, changing the
merchandising mix, improving services, but it becomes very difficult to change the location.
Therefore, recent trends have observed a lot of research being done before finalizing the location,
be it penetrating a new city, new customer segment or a new location within a city.

Retailers have a vast range of analytical techniques for their use to support their location
decision. However, in the past a great proportion of retailers have preferred instinct and personal
experience, considering the process as an ‘art’ (Simkin et al, 1985), using very subjective and
emotional methods. Most locations were fixed based on rules of thumb and intuition (Rogers,
1987).

However, over the last two decades, retailers have faced increasing pressure to adopt more
rigorous and holistic approaches (Bowlby et al, 1984) due to tightening of planning controls on
new development. The availability of computing data and capacity, and rapid reduction in cost of
IT software and hardware has spurred an immense growth in the volume of data available for
analysis (both primary and secondary) (Hernandez et al, 1995). The methods for retail location
decision making that have been widely used can be broadly classified into comparative,
predictive and knowledge based. These methods have been described further in this report.

This particular study uses an integrated Decision Making Trial and Evaluation (DEMATEL) -
Analytical Network Process (ANP) model. After establishing the fundamental criteria required
for selecting a location, DEMATEL is used to determine their effects on each other. Then, ANP
is used to assign weights to each criterion and eventually, to the alternative locations being
considered.

6
2. LITERATURE REVIEW
2.1. Existing methods and strategies

In the retail market, extensive research has been done to establish different methods for deciding
the location. These methods can be classified into three categories (Ciari et al, 2008):

 Comparative: Rules of thumb, analogues, checklist, ratio

 Predictive: Discriminant analysis, multiple regression, gravity models, cluster analysis
 Knowledge-based: Neural networks, Expert systems

Descriptions of these methods are as follows (Birkin et al, 2002):

o Rules of thumb uses experience as its main criterion. It entails intuitive and subjective
guidelines based on knowledge of the sector and company.
o Checklist entails a list of variables thought to influence store performance, with a variable
points system.
o Ratio and market support focuses on population and spending potential in the area along
with a list of potential competitors.
o Analogue methods are amongst the most popular methods still used. The principle behind
it is to forecast sales by comparing with existing stores of the same retailer located in
similar areas.
o Statistical models are another popular set of methods. They are best applied to highly
segmented markets. Linear multiple regression is the most common. This model assumes
that future sales is influenced directly by a set of variables which can be evaluated using
an equation of the type:
Y = k + c1X1 + c2X2 + …... + cnXn
The coefficients are calculated by using sales levels and attributes of existing stores.
These methods help in reducing the degree of subjectivity in decision making. However,
one drawback is the possibility of excluding some relevant factors.
o Gravity models use spatial data as the key factor in determining the attractiveness of a
location. Sales are forecasted by considering distances between competing facilities and
population densities. The major drawback of this model is that it does not consider
population segmentation.

7
 o Expert systems and neural networks use computer programs that have initial knowledge
fed in and are programmed to ‘learn’ and autonomously suggest solutions for the location
problem. However, these methods require complex algorithms and are not widely seen in
practice.

Certain spatial strategies (Laualajainen, 1987) have been observed being practiced by retail
chains. They can be classified as contagion and hierarchical strategies. The contagion strategy
entails expansion from areas that have existing stores to areas in its vicinity. Wal-mart follows
this strategy in US (Holmes, 2008). On the other hand, hierarchical strategy entails colonizing
bigger areas first, saturating the market there, and then moving to smaller areas following a
hierarchical path, disregarding locations of existing stores.

Another classification of strategies can be done on the basis of competition (Brown, 1992).
These strategies are avoidance, confrontation and predation. Avoidance entails choosing a
location as far as possible from competing stores. Confrontation entails choosing a site as near as
possible to competitors. This strategy becomes more profitable when the goods sold are not very
differentiated (Hotelling, 1992). Predation entails opening stores in spaces left unoccupied by
competitors close to them and trying to predate their customers with price strategies.

Numerous papers that have employed different models and tools to support retail location
decision have been published. The table below summarizes the papers that have been referred:

S. no. Paper title Description Author(s)

1. The retail location decision
process using GIS and the
analytical hierarchy process
This paper develops a combined Roig-Tierno et
geographic information systems al (2013)
(GIS) and analytical hierarchy
process (AHP) model for selecting a
retail location. This process was
applied for a new supermarket in
Murcia, Spain

8
2. Enhancing retail location
decision support: The
development and application
of geovisualization
3. Retail store location selection
problem with multiple
analytical hierarchy process of
decision making an application the managers of the company use
in Turkey
4. A knowledge-based system for
developing retail location
strategies
5. Spatial statistics methods in
retail location research: A case
study of Ankara, Turkey
Geovisualization refers to the visual Hernandez
exploration, synthesis, analysis and (2007)
presentation of geospatial data. This
process is presented as a better
alternative to the conventional GIS
method which has a limited ability
to handle spatial-temporal data and
visualize change. The developed
method has been applied on a
number of examples.
This paper used AHP as the Erbiyik et al
selection model. The model was (2012)
applied in Aygin Mil Inc. It helped
both, a qualitative and quantitative
approach.
This paper develops a knowledge- Arentze et al
based system using a decision table (2000)
exhaustive and exclusive of decision
rules used by Dutch retailers and
planners. This system can be used
interactively in combination with
existing spatial models for
predicting retail system operations.
This system was applied to a daily
good facility.
This paper uses Spatial auto- Ozuduru et al
regression (SAR) and (2011)
Geographically weighted regression
(GWR) to establish relationships
between shopping centre attributes
9

6. Integration of integer
programming with GIS
analyzing abilities for
determining the convenience
levels of retail stores
7. Assessing the impact of retail
location on store performance:
A comparison of Wal-Mart
and Kmart stores in Cincinnati
8. The importance of retail brand
equity and store accessibility
for store loyalty in local
competition
9. A multi-objective retail service This paper constructs a multi-
location model for fast food
and Trade area characteristics in
Ankara (Turkey).
This paper measures the Onden et al
attractiveness levels of an area for a (2012)
new retail store using spatial
closeness and density analysis.
Network analysis is used to
determine the total coverage of
attractiveness levels of each
alternative. Integer programming is
then used to integrate the spatial and
network analysis tools.
This paper analyzes and explains Li et al (2012)
differences in performances of the
two stores in Greater Cincinnati area
from a locational perspective. A
modified Huff model is which takes
competition and agglomeration into
account. It concludes that selection
of locations influences the
performance.
This paper evaluates the relative Swoboda et al
importance of store accessibility and (2013)
retail brand equity for store loyalty
based on 4151 interviews and
objective data on 30 stores. It also
analyses the negative influence of
the competitor’s brand on store
loyalty.
Min (1987)
objective retail location decision
10
 restaurants
10. Location decision making in
retail firms: Evolution and
Challenge
11. Literature review on selection
criteria of store location based
on performance measures
Table 1: Existing methods for retail location decision
model which considers the spatial
and behavioral aspects of location
scenarios, and explains the benefit
of systematic sequential decision
process. This model was then used
on real fast food restaurant location
problems.
This paper employs a questionnaire Reynolds et al
survey of retailers to analyze the (2010)
techniques used for location
assessment, 10 years since a similar
study was conducted by Hernandez
et al (2000). The study observes a
major shift to data and analytical
models from intuition and
experience.
This paper shows the results of a Turhan et al
literature review for assessing the (2013)
different selection criteria of retail
store locations. Eventually, these
criteria were classified and a
theoretical model for future research
was constructed.
11
2.2. Application of DEMATEL and ANP methods

The integrated DEMATEL-ANP method has not been used for retail location decision making
before now. However, these methods have been used to solve problems in various other fields.
The table below summarizes papers that have been referred to gauge the potential of these
methods:

S. no. Paper title Description Author(s)

1. Intuitionistic fuzzy based
DEMATEL method for
developing green practices and
performances in a green supply practices to find the main practices
chain
2. Using DEMATEL method to
analyze the causal relations on
technological innovation
capability evaluation factors in
Thai technology-based firms
3. Using DEMATEL and ANP
methods to develop a
maintenance rating program
(MRP) in Taiwan
4. Environmental site selection
for oil jetty using the analytical best geographical location for oil
network process method case
study: Boushehr, Iran
This paper uses the DEMATEL Govindan et al
method to evaluate green supply (2015)
chain management (GSCM)
to improve both economic and
environmental performances. A
case study has been performed.
This paper evaluates the technology Sumrit et al
innovation capabilities (TICs) (2012)
evaluation factors of enterprises by
using the DEMATEL method. It
determined the importance of the
criteria involved and the causal
relations among them.
This paper developed a Chang (2011)
maintenance rating program (MRP)
in Taiwan by combining
DEMATEL and ANP methods to
evaluate maintenance processes.
This paper uses ANP to choose the Hasanzadeh et
al (2014)
jetty from four spatial alternatives.
Observations were physically made
to validate this method.

12
 5. Evaluation of options for This paper uses ANP to evaluate Nixon et al
energy recovery from alternative technologies for (2013)
municipal solid waste in India producing electricity from
using the hierarchical municipal solid waste in India. It
analytical network process found ANP to provide a successful
framework for recommending a
technology.
6. A novel hybrid model based on This paper uses the DEMATEL and Tsai et al (2010)
DEMATEL and ANP for ANP methods for selecting an
selecting cost of quality model optimal cost of quality model. As a
development result of adopting this approach, it
successfully factors in the
interdependencies between criteria.
Table 2: Previous applications of DEMATEL and ANP methods

The existing efforts clearly show the lack of a method to make retail location decisions which
accounts for interdependencies between the criteria. The integrated DEMATEL-ANP method
addresses this very issue. The successful implementation of this model in other fields like quality
management, environmental engineering etc. looks promising and is therefore applied in this
study.

2.3. Criteria used for retail location decision

The criteria can be classified into Customers, Accessibility, Competition and Costs (Zentes,
2007) (McGoldrick, 2002). The sub-criteria selected for these criteria are as follows:

Customers Accessibility Competition Costs

Disposable income Road network and Existing retail stores Rent and leasing
per capita and public transport and their cumulative terms
shopping patterns attraction, saturation
Population size Available parking Vacant area for store Staff and security
space expansion costs
Table 3: Selected criteria and sub-criteria for retail location decision

13
3. DEMATEL METHOD

Once the criteria have been established, it becomes important to understand the interrelationship
among them. To do this, the Decision Making Trial and Evaluation (DEMATEL) method has
been used.

DEMATEL method was created by BMI Institute, Geneva, between 1972 and 1976. In recent
years, it has been widely used. Abdollahi et al (2014) used it to determine the impact criteria
have on each other for supplier portfolio selection. Herat et al (2011) used it to study 9 criteria
for selecting improvement projects in the healthcare sector.

The steps in DEMATEL are as follows:

Step 1: The criteria are henceforth denoted as:

C1 – Customers
C2 – Accessibility
C3 – Competition
C4 – Costs

A 4×4 direct influence matrix Z is constructed by pair wise comparison of the 4 criteria. A
measurement scale of 0, 1, 2, 3 and 4 is used to represent no influence, low influence, medium
influence, high influence and extremely high influence respectively. The zij element denotes the
degree of impact of the ith criteria on the jth criteria

C1 C2 C3 C4
C1 0 2 3 1
C2 4 0 3 3
C3 3 1 0 1
C4 1 0 3 0
Table 4: Matrix Z

14
Step 2: Computing the normalized direct relation matrix X

X=λ.Z

Where,
1 1
𝜆 = min [ , ]
𝑚𝑎𝑥 1 ≤ 𝑖 ≤ 4 ∑𝑗=1 ⎸zij ⎸ 𝑚𝑎𝑥 1 ≤ 𝑗 ≤ 4 ∑4𝑖=1 ⎸zij ⎸
4

The matrix obtained is as follows:

0 0.2 0.3 0.1

0.4 0 0.3 0.3
X=[ ]
0.3 0.1 0 0.1
0.1 0 0.3 0

Step 3: Computing the total influence matrix T

T = X (I – X)-1

The matrix obtained is as follows:

−0.17 0.21 0.28 0.03

0.37 −0.09 0.13 0.27
T=[ ]
0.31 0.05 −0.13 0.06
0.02 −0.04 0.31 −0.02

Step 4: Computing D and R

D = [di]4×1 = ( ∑4𝑗=1 𝑡𝑖𝑗 ) 4×1 , and R = [rj]’1×4 = ( ∑4𝑖=1 𝑡𝑖𝑗 )′ 1×4

Therefore, di represents the degree of impact on other criteria and rj represents the degree of
influences from other criteria.

Criterion D R
C1 0.35 0.53
C2 0.68 0.13
C3 0.29 0.59
C4 0.27 0.34

15
Step 5: Setting a threshold value and obtaining a network relationship map

A threshold value (γ) is set to filter out the negligible effects observed in the total influence
matrix T. This value is set equal to the average of all elements in T.

γ = 0.099

Accordingly, the following network relationship map is obtained:

CUSTOMERS ACCESSIBILITY

COMPETITION COSTS

Figure 1: Network relationship map for the criteria

16
4. ANP METHOD
The Analytical Network Process (ANP) is a theory extending from the Analytical Hierarchical
Process (AHP). It uses relative measurement to evaluate composite priority ratios from
individual ratios that represent relative measurements of the influence of elements that interact
with respect to control criteria. ANP allows for interdependence among the criteria and sub-
criteria, thereby solving the shortcoming of AHP. The steps in ANP are as follows:

Step 1: Forming the network structure

The network for this problem has four levels, the top level being the decision goal itself. Next
level is the criteria level followed by the sub-criteria level. The last level represents the
alternative locations being considered.

Location selection (G)

Customers Accessibility Competition Costs

(C1) (C2) (C3) (C4)

 Disposable  Road network  Existing stores  Rent and

income and and public and cumulative leasing terms
shopping transport (C21) effect (C31) (C41)
patterns (C11)  Available  Vacant area for  Staff and
 Population size parking space store expansion security costs
(C12) (C22) (C32) (C42)

Alternative 1 (A1) Alternative 2 (A2) Alternative 3 (A3)

Figure 2: The network map

17
Step 2: Forming pair-wise comparison matrices and computing priority vector

Pair-wise comparisons are done by using the 9-point priority scale, where 1 represents equal
importance and 9 represents overwhelming dominance of row element over the column element.
A score of 1/9 represents overwhelming dominance of column element over the row element.

After forming each pair-wise comparison matrix, the consistency ratio (CR) is calculated. The
CR should be less than or equal to 0.1 for logical consistency.

First, pair-wise comparison of the criteria is done with respect to the final goal, and priority
vector computed. The resulting matrices obtained are shown below:

Goal C1 C2 C3 C4 Pr. vector

C1 1 3 6 2 0.487
C2 1/3 1 4 3 0.284
C3 1/6 1/4 1 1/3 0.064
C4 1/2 1/3 3 1 0.165
CR = 0.081

Next, the priority vectors for the sub-criteria are computed by comparing them with respect to
their respective criteria. The matrices obtained are shown below:

C1 C11 C12 Pr. vector

C11 1 2 0.667
C12 1/2 1 0.333
CR = 0

C2 C21 C22 Pr. vector

C21 1 5 0.833
C22 1/5 1 0.167
CR = 0

18
 C3 C31 C32 Pr. Vector
C31 1 7 0.875
C32 1/7 1 0.125
CR = 0

C4 C41 C42 Pr. Vector

C41 1 5 0.833
C42 1/5 1 0.167
CR = 0

Next, to obtain relative weights among the criteria, pair-wise comparison of criteria is done with
respect to a fixed criterion and priority vectors calculated. The matrices obtained are shown
below:

C1 C2 C3 Pr. Vector
C2 1 8 0.889
C3 1/8 1 0.111
CR = 0

C2 C1 C3 C4 Pr. Vector
C1 1 5 7 0.747
C3 1/5 1 1 0.134
C4 1/7 1 1 0.119
CR = 0.013

C3 C1 Pr. Vector
C1 1 1
CR = 0

19
 C4 C1 Pr. Vector
C1 1 1
CR = 0
Table 5: Pair-wise comparison matrices for the criteria and sub-criteria

Now, the pair-wise comparison of the three alternatives is done with respect to a fixed sub-
criterion, and priority vectors computed. These comparisons should ideally be done after
thorough field research by experts. The matrices obtained are shown below:

C11 A1 A2 A3 Pr. Vector

A1 1 1/3 1 0.2
A2 3 1 3 0.6
A3 1 1/3 1 0.2
CR = 0

C12 A1 A2 A3 Pr. Vector

A1 1 2 4 0.571
A2 1/2 1 2 0.286
A3 1/4 1/2 1 0.143
CR = 0

C21 A1 A2 A3 Pr. Vector

A1 1 1 1/4 0.167
A2 1 1 1/4 0.167
A3 4 4 1 0.667
CR = 0

C22 A1 A2 A3 Pr. Vector

A1 1 2 1/2 0.286
A2 1/2 1 1/4 0.143
A3 2 4 1 0.571
CR = 0

20
C31 A1 A2 A3 Pr. Vector
A1 1 1 1 0.333
A2 1 1 1 0.333
A3 1 1 1 0.333
CR = 0

C32 A1 A2 A3 Pr. Vector

A1 1 2 1/2 0.286
A2 1/2 1 1/4 0.143
A3 2 4 1 0.571
CR = 0

C41 A1 A2 A3 Pr. vector

A1 1 1 1/3 0.2
A2 1 1 1/3 0.2
A3 3 3 1 0.6
CR = 0

C42 A1 A2 A3 Pr. vector

A1 1 1 4 0.444
A2 1 1 4 0.444
A3 1/4 1/4 1 0.111
CR = 0

Table 6: Pair-wise comparison matrices for the alternatives

21
Step 3: Computing the weighted super-matrix

After all the pair-wise comparisons are done, the priority vectors are represented in a
consolidated un-weighted super-matrix SU. The matrix is shown below:

G C1 C2 C3 C4 C11 C12 C21 C22 C31 C32 C41 C42 A1 A2 A3

G 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

C1 0.487 0 0.747 1 0 0 0 0 0 0 0 0 0 0 0 0

C2 0.284 0.889 0 0 0 0 0 0 0 0 0 0 0 0 0 0

C3 0.064 0.111 0.134 0 1 0 0 0 0 0 0 0 0 0 0 0

C4 0.165 0 0.119 0 0 0 0 0 0 0 0 0 0 0 0 0

C11 0 0.667 0 0 0 0 0 0 0 0 0 0 0 0 0 0

C12 0 0.333 0 0 0 0 0 0 0 0 0 0 0 0 0 0

C21 0 0 0.833 0 0 0 0 0 0 0 0 0 0 0 0 0

C22 0 0 0.167 0 0 0 0 0 0 0 0 0 0 0 0 0

C31 0 0 0 0.875 0 0 0 0 0 0 0 0 0 0 0 0

C32 0 0 0 0.125 0 0 0 0 0 0 0 0 0 0 0 0

C41 0 0 0 0 0.833 0 0 0 0 0 0 0 0 0 0 0

C42 0 0 0 0 0.167 0 0 0 0 0 0 0 0 0 0 0

A1 0 0 0 0 0 0.2 0.571 0.167 0.286 0.333 0.286 0.2 0.444 1 0 0

A2 0 0 0 0 0 0.6 0.286 0.167 0.143 0.333 0.143 0.2 0.444 0 1 0

A3 0 0 0 0 0 0.2 0.143 0.667 0.571 0.333 0.571 0.6 0.111 0 0 1

Table 7: The un-weighted super-matrix SU

22
Next, SU is made column stochastic to obtain the weighted super-matrix SW. The matrix is shown
below:

G C1 C2 C3 C4 C11 C12 C21 C22 C31 C32 C41 C42 A1 A2 A3

G 0.5 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

C1 0.243 0 0.374 0.5 0 0 0 0 0 0 0 0 0 0 0 0

C2 0.142 0.445 0 0 0 0 0 0 0 0 0 0 0 0 0 0

C3 0.032 0.055 0.067 0 0.5 0 0 0 0 0 0 0 0 0 0 0

C4 0.083 0 0.059 0 0 0 0 0 0 0 0 0 0 0 0 0

C11 0 0.334 0 0 0 0 0 0 0 0 0 0 0 0 0 0

C12 0 0.166 0 0 0 0 0 0 0 0 0 0 0 0 0 0

C21 0 0 0.416 0 0 0 0 0 0 0 0 0 0 0 0 0

C22 0 0 0.084 0 0 0 0 0 0 0 0 0 0 0 0 0

C31 0 0 0 0.437 0 0 0 0 0 0 0 0 0 0 0 0

C32 0 0 0 0.063 0 0 0 0 0 0 0 0 0 0 0 0

C41 0 0 0 0 0.416 0 0 0 0 0 0 0 0 0 0 0

C42 0 0 0 0 0.084 0 0 0 0 0 0 0 0 0 0 0

A1 0 0 0 0 0 0.2 0.571 0.167 0.286 0.333 0.286 0.2 0.444 1 0 0

A2 0 0 0 0 0 0.6 0.286 0.167 0.143 0.333 0.143 0.2 0.444 0 1 0

A3 0 0 0 0 0 0.2 0.143 0.667 0.571 0.333 0.571 0.6 0.111 0 0 1

Table 8: The weighted super-matrix SW

23
Step 4: Computing the limit super-matrix

To obtain the limit super-matrix SL, matrix SW is raised to a sufficiently large power to reach
convergence. To do this, MATLAB was used to raise SW to the power 999999. The matrix SL is
shown below:

G C1 C2 C3 C4 C11 C12 C21 C22 C31 C32 C41 C42 A1 A2 A3

G 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

C1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

C2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

C3 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

C4 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

C11 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

C12 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

C21 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

C22 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

C31 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

C32 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

C41 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

C42 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

A1 0.270 0.283 0.236 0.305 0.273 0.2 0.571 0.167 0.286 0.333 0.286 0.2 0.444 1 0 0

A2 0.334 0.386 0.266 0.347 0.294 0.6 0.286 0.167 0.143 0.333 0.143 0.2 0.444 0 1 0

A3 0.397 0.331 0.498 0.347 0.433 0.2 0.143 0.667 0.571 0.333 0.571 0.6 0.111 0 0 1

Table 9: The limit super-matrix SL

24
5. RESULT AND CONCLUSION

The limit super-matrix SL shows the relative weights of the three alternatives with respect to the
final location decision goal. The results are tabulated below:

Alternative Weights
A1 0.270
A2 0.334
A3 0.397
Table 10: Weights for the alternatives

Accordingly, location A3 is the most preferable option and location A1 is the least.

This study clearly demonstrates how the DEMATEL-ANP method can be effectively used to
make complicated managerial decisions. It reduces the level of subjectivity involved and helps
managers to rely less on intuition when the stakes are high. Additionally, this method also factors
in the interdependent relationships of the criteria.

25
6. REFERENCES
1. Abollahi, M., Arvan, M. and Razmi, J. (2014). “An integrated approach for supplier
portfolio selection: Lean or Agile?”. Expert systems with applications, 679-690
2. Arentze, T., Borgers, A. and Timmermans, H. (2000). “A knowledge-based system for
developing retail location strategies”. Computers, Environment and urban systems,
volume 24, 489-508
3. Birkin, M., G. Clarke and M. Clarke (2002). Retail Geography and Intelligent Network
Planning, John Wiley, Chichester
4. Chang, J. (2011). “Using DEMATEL and ANP methods to develop a maintenance rating
program (MRP) in Taiwan”. IAARC, 1312-1313
5. Cheng, E. and Li, H. (2004). “Contractor selection using the analytical network process”.
Construction management and economics, 1021-1032
6. Ciari, F., Lochl, M and Axhausen, K. (2008). “Location decision of retailers: an agent-
based approach”. International conference on recent advances in retailing and services,
Zagreb
7. Erbiyik, H., Ozcan, S. and Karaboga, K. (2012). “Retail store location selection problem
with multiple analytical hierarchy process of decision making and application in Turkey”.
Procedia – Social and Behavioral sciences, volume 58, 1405-1414
8. Govindan, K., Khodaverdi, R. and Vafadarnikjoo, A. (2015). “Intuitionistic fuzzy based
DEMATEL method for developing green practices and performances in a green supply
chain”. Expert systems with applications
9. Hasanzadeh, M. and Danekhar, A. (2014). “Environmental site selection for oil jetty
using the analytical network process method case study: Boushehr, Iran”. Ocean
Engineering, volume 77, 55-60
10. Herat, A., Noorossana, R., Parsa, S. and Serkani, E. (2012). “Using DEMATEL-ANP
hybrid algorithm approach for selecting improvement projects of Iranian excellence
model in healthcare sector”. African journal of business management, 627-645
11. Hernandez, T. (2007). “Enhancing retail location decision support: The development and
application of geovisualization”. Journal of retailing and consumer services, volume 14,
249-258

26
12. Li, Y. and Liu, L. (2012). “Assessing the impact of retail location on store performance:
A comparison of Wal-Mart and Kmart stores in Cincinnati”. Applied geography, volume
32, 591-600
13. McGoldrick, P. (2002). Retail marketing, 2nd edition, McGraw-Hill, London
14. Min, H. (1987). “A multi-objective retail service location model for fast-food
restaurants”. Omega, volume 15, 429-441
15. Nixon, J., Dey, S., Ghosh, S. and Davies, P. (2013). “Evaluation of options for energy
recovery from municipal solid waste in India using the hierarchical analytical network
process”. Energy, volume 59, 215-223
16. Onden, I., Sen, C. and Sen, A. (2012). “Integration of integer programming with GIS
analyzing abilities for determining the convenience levels of retail stores”. Procedia –
Social and Behavioral sciences, volume 62, 1144-1149
17. Ozuduru, B. and Varol, C. (2011). “Spatial statistics methods in retail location research:
A case study of Ankara, Turkey”. Procedia environmental sciences, volume 7, 287-292
18. Reynolds, J. and Wood, S. (2010). “Location decision-making in retail firms: Evolution
and Challenge”. International Journal of Retail & Distribution Management, volume 38,
828-845
19. Roig-Tierno, N., Baviera-Puig A., Buitrago-Vera J. and Mas-Verdu, F. (2013). “The
retail site location decision using GIS and analytical hierarchy process”. Applied
Geography, volume 40, 191-198
20. Saaty, T. (1999). “Fundamentals of the Analytic Network Process”. International
symposium on the analytical hierarchy process, Kobe, Japan
21. Simkin, L., Doyle, P. and Saunders, J. (1985). “How retailers put site location techniques
into operation: An assessment of major multiple’s practice”. Retail and Distribution
management, volume 13, issue 3, 21-26
22. Sumrit, D. and Anuntavoranich, P. (2013). “Using DEMATEL method to understand the
causal relations on technological innovation capability evaluation factors in Thai
technology-based firms”. International transaction journal of engineering, management,
and applied sciences and technologies, volume 4, issue 2, 81-103

27
23. Swoboda, B., Berg, B., Schramm-Klein, H. and Foscht, T. (2013). “The importance of
retail brand equity and store accessibility for store loyalty in local competition”. Journal
of retailing and consumer services, volume 20, 251-262
24. Turhan, G., Akalin, M. and Zehir, C. (2013). “Literature review on selection criteria of
store location based on performance measures”. Procedia – Social and Behavioral
sciences, volume 99, 391-402
25. Wen-Hsien, T. and Wei, H. (2010). “A novel hybrid model based on DEMATEL and
ANP for selecting cost of quality model development”. Total quality management &
business excellence, volume 21, 439-456
26. Zentes, J., D. Morschett and H. Schramm-Klein (2007). Strategic Retail Management,
Gabler, Wiesbaden

28