International Journal of Computer Applications (0975 – 8887)

Volume 5– No.9, August 2010

Vendor Evaluation Using Multi Criteria Decision Making

Technique
C. Elanchezhian B. Vijaya Ramnath Dr. R. Kesavan
Research Scholar, Research Scholar, [Link],
Dept. of Production Technology, Dept. of Production Technology, Dept. of Production Technology,
[Link], Anna University, [Link], Anna University, [Link], Anna University,
Chrompet, Chennai-600 044. Chrompet, Chennai-600 044. Chrompet, Chennai-600 044.

ABSTRACT manufacturing of a product from the procurement of raw materials

A supply chain is a network of departments, which is involved in to the distribution of the final product to the customer.
the manufacture of a product from the procurement of raw The purchasing function has gained greater importance in the
materials to the distribution of the final products to the customer. supply chain management due to factors such as globalization,
The term supply chain is already invoked effervescence among the increased value addition in supply, and accelerated technological
managerial community. The purchasing function has gained change. Purchasing involves buying the raw materials, supplies
importance in the supply chain management due to factors such as and components for the organization. The activities associated
globalization, increased value addition in supply and accelerated with it include selecting and qualifying suppliers, rating supplier
technology change. A key and perhaps the most important process performance, negotiating contracts, comparing price, quality and
of the purchasing function is the efficient selection of suppliers, service, sourcing goods and service, time purchases, selling terms
because it brings significant savings for the organization. In of sale, evaluating the value received, predicting price, service and
general, the supplier selection criteria most commonly used by the sometimes demand changes, specifying the form in which goods
industries are quality, delivery and price. Also, depending on the are to be received etc. A key and perhaps the most important
corporate environment of the industries, the importance of the process of the purchasing function is the efficient selection of
performance measure can vary. In this work a versatile technique suppliers, because it brings significant saving for the organization.
namely multi criteria decision making (MCDM) technique which The objective of the supplier selection process is to reduce risk
involves the analytical network process (ANP) and technique for and maximize the total value for the buyer, and it involves
order performance by similarity to idea solution (TOPSIS) method considering a series of strategic variables.
has been used to select the best vendor.
Some authors have identified several criteria for supplier
Analytical Network Process and TOPSIS method are powerful selection, such as the net price, quality, delivery, historical
decision making processes which help people to set priorities on supplier performance, capacity, communication systems; service
parameters that are to be considered by reducing complex an geographic location among others. These criteria are a key
decision to a series of one-to-one comparisons, thereby issue in the supplier assessment process since it measures the
synthesizing the result performance of the suppliers.
When any vendor for a particular item make changes for the
parameters like price, quality etc to improve his performance or 2. LITERATURE SURVEY
has improved abilities in managing supply chain by providing Satu Peltola et al(2002) emphasized the use of Analytical
better delivery to his customer, the whole hierarchy process for Hierarchy Process(AHP) and integrated Support System(GSS)
arriving the ranking of vendors is to be performed again for method for improving the business performance in every sector.
finding out the best vendor. Now it is felt that a standard In this paper, the performance of buyer-supplier relationship will
automated procedure which could perform the above processing be enhanced by using a benchmarking method in aiding the
task is essential. So, standard software was developed in a suitable identification and implementation of development actions
platform such as VB, .NET and MS access that could meet the required for reaching the world-class level. The benchmarking
current requirement. This package can be executed several process will be led by two decision support systems namely AHP
numbers of times with changing input parameters values thus and GSS, to ensure valuable outcomes of the benchmarking with
serving the purpose. fewer resources
Jiann Liang Yang et al(2008) proposed an integrated fuzzy
1. INTRODUCTION multiple criteria decision making(MCDM) techniques for solving
Traditionally organization have been divided into operative vendor selection problem, Jiann utilize triangular fuzzy number to
functions such as marketing, planning, purchasing, finance etc. express the subjective preference of evaluators with respect to the
supply chain is a strategy that integrates these functions creating a considered criteria
general plan for organization which satisfies the service policy,
maintaining the lowest possible cost level due to the incredible Banar et al(2006) used analytical network process(ANP), one of
competitive environment that they are exposed to. A supply chain the multi-criteria decision making (MCDM) tools to choose one
is a network of departments, which is involved in the of the four alternative landfill sites for the city of Eskisehir,
Turkey. For this purpose super decision software has been used

4
 International Journal of Computer Applications (0975 – 8887)
Volume 5– No.9, August 2010

and benefit opportunity cost and risk(BOCR) analysis has been 2. Purchasing manager
done to apply ANP. 3. Quality manager
4. Product manager
Cevriye Gencer and Didem Gurpinar(2007) used Analytical
5. Production manager
Network Process(ANP) for selecting suppliers in an electronic
industry . suppliers selection, which is the first step of the
activities in the product realization process starting from the 4. PROPOSED METHODOLOGY
purchasing of raw material till the end of delivering the products, 4.1 Steps involved in the Proposed Model
is evaluated as a critical factor for the companies desiring to be The selection process has been modified to a five-step hybrid
successful in today’s competitive condition. With the scope of this procedure, as follows
paper, suppliers selection was considered as a multi criteria
decision problem. Step 1: Identification of necessary criteria for vendor selection
Step 2: Recognition of the independence between criteria
Hsu Shih et al(2006) used an extension of TOPSIS( Technique for Step 3: Calculating the weights of criteria
Order Performance by Similarity to Ideal Solution), a multi- Step 4: Evaluation of vendors
attribute decision making(MADM) technique, to a group decision Step 5: Negotiation for the purchase
environment. TOPSIS is a practical and useful technique for
ranking and selection of a number of externally determined
alternatives through distant measures.
Ching Torng Lin et al(2006) developed a fuzzy agility index(FAI)
based on agility providers using fuzzy logic. To achieve a
competitive edge in the rapidly changing business environment,
companies must align with suppliers and customers to streamline
operation, as well as working together to achieve a level of agility
beyond individual companies. To illustrate the efficacy of the
method, the study also evaluates the supply chain agility of a
Taiwanese company.
Desheng Wu et al(2008) considered three types of risk evaluation
models within supply chain such as chance constrained
programming(CCP), data envelopment analysis(DEA), and multi-
objective programming(MOP) models. Various risks are modeled
in the form of probability and simulation of specific probability
distribution in risk-embedded attributed is conducted in these
three types of risk evaluation models
Filip Roodhooft and Jozef Konings(1996) proposed an activity Figure 1: The proposed framework for vendor selection
based costing approach for vendor selection and evaluation. This
system allows us to compute total costs caused by a supplier in a 4.2 Evaluation Criteria with Interdependence
firm’s production process, thereby increasing the objectivity in the In the vendor evaluation process, an objective, unbiased
selection process. decision is very hard to reach given the numerous criteria that
need to be carefully considered and examined. One formal group
3. MEASURING SUPPLIER management technique for determining a set of evaluation criteria
is NGT. This well-known process forces everyone to participate
PERFORMANCE and no dominant person is allowed to come out and control the
Need for measuring supplier performance proceedings. In NGT, all ideas have equal stature and will be
Supplier performance has to be measured occasionally for the judged impartially by the group. In our problem, seven potential
following reasons evaluation criteria are determined as follows

1. To increase performance visibility 1. On-time delivery (Cl).

2. To uncover and remove hidden waste and cost drivers in the 2. Product quality (C2).
supply chain 3. Price/cost (C3).
3. To leverage the supply base 4. Facility and technology (C4).
4. To align customer and supplier business practices 5. Responsiveness to customer needs (C5).
5. To mitigate risk 6. Professionalism of salesperson (C6).
6. To improve supplier performance 7. Quality of relationship with vendor (C7).
To simplify the process and avoid any misunderstandings, the
interaction between any two of these criteria is not considered in
THE DECISION MAKING TEAM the first instance. These criteria may not include all of the
The decision making team comprises of the following members decision factors in vendor selection. However, they are indeed
meaningful measures and have been emphasized in many leading
1. Purchasing director articles

5
 International Journal of Computer Applications (0975 – 8887)
Volume 5– No.9, August 2010

Next, in order to reflect the interdependence property between the Next, the effects of the interdependence between the criteria are
criteria, we need to identify the exact relationship in a network resolved. The group members will examine the impact of all
structure of ANP. Another NGT process is taken to construct the criteria on each other by pair-wise comparisons too. To help
relationship based on the following two recognitions: smooth the comparison process, a couple of questions such as
"Which criterion will influence criterion C3 more: C2 or C7? And
1. Price/cost may be influenced by the quality of products and
how much more?" are answered. Various pair-wise comparison
the relationship with vendors.
matrices are constructed for each criterion. These pair-wise
2. Product quality may be influenced by facility and comparison matrices are needed for identifying the relative
technology. impacts of criteria interdependent relationships. The normalized
principal eigen vectors for these matrices are calculated and
Figure 1.2 represents the relationship of interdependency. A shown as column components in interdependence weight matrix
single arrow implies a one-way relationship. For example, the B, where zeros are assigned to the eigenvector weights of the
arrow that leaves from C2 and feeds into C3 implies that the criteria with no interdependent relationship.
relationship of criterion C2 has an influence on criterion C3.
ACTIVITY: 3
4.3 Determination of Weights to Criteria Now the interdependence priorities .of the criteria can be obtained
To determine the relationship of the degree of interdependence,
by synthesizing the results from the previous two steps as follows:
the ANP technique, which is an extension of AHP, is used to
address the relative importance of the criteria. ANP is developed
to generate priorities for decisions without making assumptions
wc Bw T2 …. (1)
about a unidirectional hierarchy relationship between decision
Thus, the weights of the evaluation criteria can be determined
levels. To take the place of a linear top-to-bottom form of strict
hierarchy, the ANP model provides a looser network structure and 4.4 The Ranking and Selection Process
possibly represents any decision problem. The relative importance As large number of potential available vendors in the current
or strength of the impacts qn a given element is measured on a marketing environment, a full ANP decision process becomes
ratio scale, which is similar to AHP. impractical in some cases. To avoid an unreasonably large number
of pair-wise comparisons, we choose TOPSIS as the ranking
technique because of its concept's ease of use. Also, ANP is
adopted simply for the acquisition of the weights of criteria. First,
a general TOPSIS process with six activities is listed below.
ACTIVITY: 1
Establish a decision matrix for the ranking.. The structure of
the matrix can be expressed as follows:
F1 F2  Fj  Fn
A1 f11 f12  f1 j  f1n
Figure 2: The interdependent relationship between the selected criteria A 2 f 21 f 22  f 2 j  f 2n
ACTIVITY: 1 D       
Without assuming the interdependence between criteria, the A i f i1 fi2  f ij  f in
decision makers or experts are asked to evaluate all proposed
criteria pair-wise. They responded to questions such as "Which       
criteria should be emphasized more in a vendor, and how much
more?" The responses were presented numerically and scaled on A m f m1 f m 2  f m j  f m n
the basis of Saaty's 1- 9 scale, where 1 represents indifference
between the two criteria and 9 represents extreme preference for where Ai denotes the alternatives i,j = 1,……m; Fj represents jth
one criterion over the compared criterion. Each pair of criteria is attribute or criterion, j=1,….., n related to ith alternative; and fij is
judged only once. A reciprocal value will be assigned a crisp value indicating the performance rating of each alternative
automatically for the reverse comparison. Once the pair-wise Ai with respect to each criterion Fj.
comparisons are completed, the local priority vector wx is ACTIVITY: 2
computed as the unique solution of
Calculate the normalized decision matrix R (=[rij]). The
Aw1 m axw1 . normalized value rij is calculated as shown in Eq. (2).

Where max is the largest eigen value of pair-wise comparison f ij

matrix A. All obtained vectors are further normalized to represent rij …. (2)
n
the local priority vector w1. 2
f ij
ACTIVITY: 2 i 1

6
 International Journal of Computer Applications (0975 – 8887)
Volume 5– No.9, August 2010

where j = 1,….., n; i =1,……., m. n

2
ACTIVITY: 3 Di rij rj wj , i 1.......m.
j 1
Calculate the weighted normalized decision matrix by multiplying
the normalized decision matrix by its associated weights. The
weighted normalized value vij is calculated as shown in Eq. (3). D i
Ci , i 1, .......,m …(9)
v ij w j rij , j = 1,….., n; i =1,……., m. …. (3) D i D i

where Wj represents the weight of the jth attribute or criterion.

5. RESULTS AND DISCUSSION
ACTIVITY: 4 For illustration purpose, an example of vendor selection is
Determine the PIS and NIS, respectively: performed by the suggested hybrid approach. Let us consider that
four vendors, Al, A2, A3, and A4 are involved for evaluation. A
v v1 .......... ... v n team of three is charged in this project. Seven criterias are
considered for the selection (Step 1 and Step 2), and the other
steps are summarized as follows:
max vij | j J , min vij | j J '
Step 3: The decision makers will be asked to evaluate all criteria
pair-wise without assuming the interdependence between them.
v v1 .......... ... v n The normalized eigen vector can be calculated as w2 = (Cl, C2,
C3, C4, C5, C6, C7) = (0.347, 0.247, 0.142, 0.035, 0.084, 0.043,
min vij | j J , max vij | j J ' 0.101) which represents the related local priority of these criteria.
The degree of consistency of the pair-wise comparison is
where J is associated with the benefit criteria, and j is associated measured with the use of the consistency ratio (CR) index. It is
with the cost criteria. considered logically consistent if CR is less than or equal to 0.1.
ACTIVITY: 5 The CR value for this case is 0.058, which is acceptable.

Calculate the separation measures, using the m-dimensional Step 4: The interdependence between the criteria is now
Euclidean distance. The separation measure Di+ of each considered. All decision makers or group members will examine
alternative from the PIS is given in Eq. (4). the impact of all the criteria by pair-wise comparison. In total,
there are seven comparison matrices generated by all members.
n The normalized eigenvector for these matrices developed by the
2
Di vij v j , i 1.......m. …. (4) first member is calculated and shown as seven columns in Table
j 1
2, where zeros are assigned to the eigenvector weights of the
criteria with no interdependent relationship. The data in Table 2
Similarly, the separation measure Di– of each alternative from the imply the relative impact of part of the criteria on others. For
NIS is as shown in Eq. (5). example, the degree of relative impact of C2 for C3 is 0.236.

n
2
5.1 Pair-wise Comparison Matrix for Criteria
Di v ij v j , i 1.......m. …. (5) The pair-wise comparison matrix for different criterias is shown in
j 1 Table 1.
Table 1 The pair-wise comparison matrix for criteria
ACTIVITY: 6
Calculate the relative closeness to the idea solution and rank the Vector
C1 C2 C3 C4 C5 C6 C7
alternatives in descending order. The relative closeness of the weights
alternative Ai with respect to PIS V+ can be expressed as:
C1 1 2 4 5 4 6 4 0.346
Di C2 ½ 1 3 4 4 5 3 0.247
Ci , i 1, .......,m …. (6)
Di Di C3 ¼ 1/3 1 1/3 2 1/5 1/2 0.142

Where the index value of C i lies between 0 and 1. the larger the C4 ¼ 1/5 3 1 4 1/2 2 0.035
index value, the better the performance of the alternatives. C5 ¼ 1/4 1/2 1/4 1 1/4 1/3 0.084
R+ = { r1+ …..r n+} 1/6 1/5 5 2 4 1 3 0.043
C6
…(7)
C7 ¼ 1/3 2 1/2 3 1/3 1 0.101
n
2
D rij rj wj , i 1.......m. …(8)
j 1

7
 International Journal of Computer Applications (0975 – 8887)
Volume 5– No.9, August 2010

5.2 Degree of Relative Impact for Evaluation 5.3 Normalized Decision Matrix
Criteria The normalized decision matrix of various vendors with respect to
Table 2 represents the degree of relative impact for evaluation the criterias is shown below in Table 3.
criteria. Table 3 Degree of relative impact for evaluation criteria
Table 2 Degree of relative impact for evaluation criteria
C1 C2 C3 C4 C5 C6 C7
C1 C2 C3 C4 C5 C6 C7 A1 0.552 0.396 0.431 0.453 0.462 0.629 0.375
C1 1 0 0 0 0 0 0 A2 0.552 0.594 0.323 0.543 0.577 0.449 0.375
C2 0 0.866 0.236 0 0 0 0 A3 0.442 0.495 0.647 0.543 0.577 0.449 0.600
C3 0 0 0.606 0 0 0 0 A4 0.442 0.495 0.539 0.453 0.346 0.449 0.600
C4 0 0.134 0 1 0 0 0 Step 5: based on the PIS and N, the ranking activities will start.
By Eq. (7), the PIS and NIS for decision maker 1 will be:
C5 0 0 0 0 1 0 0
R+= (0.552, 0.594, 0.647, 0.543, 0.577, 0.629, 0.600),
C6 0 0 0 0 0 1 0
R+= (0.442, 0.396, 0.323, 0.453, 0.346, 0.449, 0.375),
C7 0 0 0.158 0 0 0 1
The relative importance of the criteria considering
5.4 Separation distances of the Group
Table 4 shows the separation distances calculated by the three
interdependence can be obtained by synthesizing the results.
decision makers based on the performance of vendors for the
chosen criterias.
C1 1 0 0 0 0 0 0
Table 4 Separation distances of the group
C2 0 0.866 0.236 0 0 0 0
Aggregated
C3 0 0 0.606 0 0 0 0 DM#1 DM#2 DM#3 separation
wi C4 0 0.134 0 1 0 0 0 distances

C5 0 0 0 0 1 0 0 Di Di Di Di Di Di Di Di
C6 0 0 0 0 0 1 0 A1 0.151 0.064 0.208 0.080 0.236 0.017 0.195 0.044
C7 0 0 0.158 0 0 0 1 A2 0.146 0.120 0.185 0.11 0.141 0.165 0.157 0.131
0.346 0.346 A3 0.089 0.165 0.126 0.234 0.025 0.222 0.066 0.205

0.247 0.247 A4 0.120 0.119 0.197 0.133 0.160 0.103 0.156 0.118

0.142 0.086 Note: DM - decision maker.

0.035 0.068 5.5 Final Rank of the Vendor Selection

Problem
0.084 0.084 Table 5 shows the closeness coefficient of various vendors which
0.043 0.043 is calculated by TOPSIS method. It ranges between 0 and 1.
Table 5.5 Final rank of the vendor selection problem
0.101 0.123
Closeness
According to the vector from decision maker 1, C1, C2, and C7 Rank Alternative
coefficient
are three of the most important factors related to the evaluation
process. 1 A3 0.758
At the next level of the decision process, the decision makers will 2 A2 0.457
be asked to establish the decision matrix by comparing candidates
under each criterion separately. The criteria are assumed to be 3 A4 0.430
benefit criteria and they were asked to give a set of crisp values
within the range of 1 to 10 to represent the performance of each 4 A1 0.185
alternative with respect to each criterion. After the decision Using the criteria weights (Wc) obtained from step 1-3 and Eq.
matrices are determined, we normalize these matrices via Eq. (2). (8), the weighted Euclidean distances, between At and R+, and
Table 3 shows the result of decision maker between Ai and R , can be calculated immediately. Table 4

8
 International Journal of Computer Applications (0975 – 8887)
Volume 5– No.9, August 2010

represents the separation distances developed by all three selection', International journal of Physical Distribution &
members. Next, to derive group priorities, the group’s aggregated Logistics Management, 30 No. 2, 96-111.
separation distances are generated by its geometric mean. The last [5] Cevriye Gencer and Didem Gurpinar (2007), 'Analytical
two columns of Table 4 show the results. Finally, the relative Network Process in supplier selection: A case study in an
closeness to the idea solution of each alternative can be calculated electronic firm', Applied Mathematical Modelling, 31, 2475-
using Eq. (9). The final results can be seen in Table 5. According 2486.
to the closeness coefficient, the ranking order of the four
candidates is A3, A2, A4 and A1. Obviously, the best selection is [6] Ching-Torng Lin, Hero Chiu and Po-Young Chu (2006),
candidate A3. 'Agility index in the supply chain', International Journal of
Production Economics, 100, 285-299.
6. SOFTWARE DEVELOPMENT [7] Chopra S. and Meindle P. (2001), 'Supply Chain
In order to execute several number of times with changing input Management Strategy, Planning and Operation', Prentice
parameters, software to calculate vendor rating is being developed Hall Inc., Upper Saddle River, 1-24.
during the phase II using [Link] and MS Access Query [8] Degraeve Z., Labro E. and Roodhooft F. (2000), 'An
Language evaluation of vendor selection models from a total cost of
ownership perspective', European Journal Of Operational
7. CONCLUSION Research, 125, 34-58
Developing a robust, easy-to-deploy method of evaluating [9] Desheng Wu and David L. Olson (2008), 'Supply chain risk,
suppliers is a-critical business competency. The methodology simulation, and vendor selection', International Journal of
should be sound and the approach practical. Gathering data for the Production Economics, 114, 646-655.
sake of data will not produce the return on investment in supplier [10] Dickson G.W. (1966), 'An analysis of vendor selection
evaluation. Most importantly, companies need to use the results as systems and decisions', Journal of Purchasing, 2, No. 1, 5-17.
a means to foster communications and a starting point for supplier
development and performance improvement. This, in turn, will [11] Dilay Celebi and Demet Bayraktar (2008), 'An integrated
help companies to reap the financial and competitive rewards of neural network and data envelopment analysis for supplier
high performing key suppliers. In this project an attempt has been evaluation under incomplete information', Expert Systems
made to select the best vendor by using Multi Criteria Decision with Applications, 35, 1698-1710.
Making technique which uses Analytical Network process (ANP) [12] Elena Tsiporkova and Veselka Boeva (2006) 'Multi-step
and TOPSIS method. Analytical Network Process mainly avoids ranking of alternatives in a multi-criteria and multi-expert
arbitrary assignments of weights for the factors depending on the decision making environment', International Journal of
decision maker, and relative pair wise comparison still makes the Information Sciences, 176, 2673-2697.
process more feasible and more accurate vendor rating can be [13] Filip Roodhooft and Jozef Konings (1996),'Vendor selection
obtained when compared to any other method. A software and evaluation - An activity based costing approach',
program has been developed during phase II by using [Link] European Journal of Operational Research, 96, 97-102.
and Ms Access by considering various factors such as quality,
price, on-time delivery, responsiveness, etc. as input to evaluate [14] Hsu-Shih Shih, Huan-Jyh Shyur and Stanley Lee E. (2006),
the best vendor as an output. In this process time, manpower 'An extension of TOPSIS for group decision making',
requirement is very less as it is based on a software and no Mathematical and Computer Modelling, 45, 801-813.
technical knowledge is required to operate this which can be done [15] Huan Jyh Shyur and Hsu-Shih Shih (2006), 'A hybrid
by a simple data entry operator. MCDM model for strategic vendor selection', Mathematical
and Computer Modelling, 44, 749-761.
8. REFERENCES [16] Jiann Liang Yang, Huan Neng Chiu, Gwo Hshiung Tzeng
[1] Alberto De Toni and Guido Nassimbeni (2001), 'A method and Ruey Huei Yeh (2008), 'Vendor selection by integrated
for the evaluation of suppliers co-design effort', International fuzzy MCDM techniques with independent and
Journal of Production Economics, 72, 169-180. interdependent relationships', International Journal of
[2] Amir Sanayei, Farid Mousavi S., Abdi M.R. and Ali Information Sciences, 178, 623-642.
Mohaghar (2008), 'An integrated group decision-making [17] Mitsutoshi Kojima, Kenichi Nakashima and Katsuhisa Ohno
process for supplier selection and order allocation using (2008), 'Performance evaluation of SCM in JIT
multi-attribute utility theory and linear programming', environment', International Journal Of Production
Journal of the Franklin Institute, 16, 823-847. Economics, 115, 439-443.
[3] Banar M, Kose B.M. and Ozkan A. (2006), 'Choosing a [18] Mohammad Taghi Taghavifarad and Daniel Mirheydari
municipal landfill site by Analytical Network Process', (2008),'A new framework for evaluation and prioritization of
Article of Environ Geol, 52, 741-751. supplier's using a Hierarchy Fuzzy TOPSIS', Proceedings of
[4] Bragila M. and Petroni A. (2000), 'A quality assurance – world academy of science, 31, 1-18.
oriented methodology for handling trade-offs in supplier