See discussions, stats, and author profiles for this publication at: [Link]

net/publication/282640046

Supplier selection using artificial neural network and genetic algorithm

Article  in  International Journal of Indian Culture and Business Management · October 2015

DOI: 10.1504/IJICBM.2015.072428

CITATIONS READS

10 2,335

1 author:

Manish Gupta
Motilal Nehru National Institute of Technology
33 PUBLICATIONS   132 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

smart supply chain View project

e procurement View project

All content following this page was uploaded by Manish Gupta on 07 October 2015.

The user has requested enhancement of the downloaded file.

Int. J. Indian Culture and Business Management, Vol. 11, No. 4, 2015 457

Supplier selection using artificial neural network and

genetic algorithm

Nitesh Asthana* and Manish Gupta

Department of Mechanical Engineering,
Motilal Nehru National Institute of Technology,
Allahabad 211004, India
Email: niteshasthana2008@[Link]
Email: mgupta@[Link]
*Corresponding author

Abstract: This paper presents the integration of GA with ANN to develop a

decision making model for the selection of the best supplier. This work
integrates multi-attribute decision making models that give grades to suppliers
on a set of criteria with the mathematical programming techniques that model
the constraints and an objective function to select the best supplier. There exists
a number of criteria for supplier selection and in this work based on the
preliminary field visits some of the criteria are chosen for supplier selections
that are more dominant in the Indian scenario. These selected criteria are:
quality, delay time, unit cost, quantity and service. Supplier score was
calculated based on the data collected from the reputed automobile industry.
This supplier score was further used for ranking the suppliers among the
number of options available.

Keywords: supplier selection; genetic algorithm; artificial neural network;

ANN; decision making model; optimisation; multi-criteria decision making;
MCDM.

Reference to this paper should be made as follows: Asthana, N. and Gupta, M.

(2015) ‘Supplier selection using artificial neural network and genetic
algorithm’, Int. J. Indian Culture and Business Management, Vol. 11, No. 4,
pp.457–472.

Biographical notes: Nitesh Asthana is a Research Scholar in the Department

of Mechanical Engineering, MNNIT, Allahabad. His current area of research
includes operation management.

Manish Gupta is working as an Assistant Professor in the Department of

Mechanical Engineering, MNNIT, Allahabad, and has more than 12 years of
experience in teaching. His current areas of research include computer-aided
manufacturing, supply chain management, electronic procurement and
industrial engineering.

Copyright © 2015 Inderscience Enterprises Ltd.

458 N. Asthana and M. Gupta

1 Introduction

Supplier selection and supplier evaluation is one the most important activity in the
management of supply chain. Selection and evaluation of suppliers is one of the most
critical activities of a company and a strategic purchasing decision that commits
significant resources (about 40–80% of total product cost) and impacts the total
performance of the firm (Gupta and Narain, 2012). The global competition among
industries to respond quickly to customers has led to the selection of the best supplier
from the lot. In many industries the cost of raw materials and components constitute a
major portion of the product cost so the supplier selection decisions determine how many
and which suppliers should be selected, selection is said to be efficient when we obtain
not only a desirable solution but also optimal solution (Arunkumar et al., 2007). Demirtas
and Ustun (2008) have suggested two kinds of supplier selection problems,

1 one supplier can satisfy all the buyer’s needs (single sourcing)

2 no single supplier can satisfy all the buyer’s/customer’s requirement (multiple

sourcing).

In such circumstances, management wants to split order quantities among suppliers for a
number of reasons, which includes creating a constant environment of competitiveness
(Gupta and Narain, 2012). The success of the supply chain is highly dependent on the
selection of good suppliers, simply looking for suppliers that is offering lowest price is
not efficient sourcing any more, multiple criteria are needed to be taken into account
when selecting suppliers (Ng, 2008).
Defining the criteria’s for selection of the supplier is also an important task. Early in
1960s, Dickson identified 23 criteria’s that ought to be considered by purchasing
personnel in evaluating suppliers (Ng, 2008). These factors are both qualitative and
quantitative. The main criteria for supplier and key partner selection are: quality,
delivery, performance history, production capability, service, engineering and technical
capability, business structure, price, integrity, warranties, honesty, reliability, reputation,
commitment and financial position (Dickson, 1966; Tangpong et al., 2010; Pal and
Kumar, 2008; Levary, 2007; Gulen, 2007). Some of these factors are linked with each
other, whereas some contradict others.
Supplier selection process often involves the selection of certain criteria while
eliminating the other criteria. Consider the state in which one supplier is providing goods
at cheaper rates but is not able to deliver on time also the quality of goods/parts is not
good. On the other hand another supplier provides goods of better quality but its price is
high also delivery time is more. Therefore, supplier selection belongs to the class of
multi-criteria decision making (MCDM) problem in which the firms need to identify the
top priorities of selecting the best supplier based on its working style and the industry
type (Agarwal et al., 2011). There exists many MCDM techniques for supplier selection
some of them are analytic hierarchy process (AHP), analytic network process (ANP),
case-based reasoning (CBR), data envelopment analysis (DEA), fuzzy set theory, genetic
algorithm (GA), artificial neural network (ANN), mathematical programming, simple
multi attribute rating techniques (SMART), and their hybrids (Dey et al., 2010). These
approaches are used individually or are integrated together to solve supplier selection
problems.
 Supplier selection using artificial neural network and genetic algorithm 459

In this paper, we have developed a decision making model by integrating GA with

ANN to select the best and optimal supplier. The main aim is to integrate, multi-attribute
decision making models that give grades to suppliers on a set of criteria with the
mathematical programming techniques that model the constraints and an objective
function to select the optimal supplier. ANN has been widely used for modelling complex
process due to their learning, accommodation of nonlinear variables, generalisation
capabilities, adaptability to change environments and resistance the missing data (Azeem
and Paul, 2011). GA is used because it is an evolutionary algorithm (EA)/optimisation
technique based on Darwin’s theory i.e. ‘survival of fittest’. This decision making model
consist of two parts: Part I applies GA to find the optimal value of function of quality,
delay time, unit cost, quantity, service. Part II uses these optimal values as an input to
ANN to find the supplier score. In this paper, Section 2 deals with literature review,
Section 3 with methodology, Section 4 with calculations, Section 5 with model validation
and Section 6 and Section 7 with discussion and conclusion respectively.

2 Literature review

Supplier selection an important element in supply chain management, has been gaining
attention in both academic literature and industrial practice (Kouvelis et al., 2004). The
role of a manufacturer, in the past decade has evolved from the manufacturer to one that
coordinates or manages the supply chain entities (Choy and Lee, 2003). This greatly
increases a firm’s dependency on suppliers which, in turn, increase the need for effective
supplier or partner selection. Narasimhan et al. (2001) states “supplier evaluation is a
complex process that involves the consideration of numerous factors”. Talluri and
Narasimhan (2004) emphasise that managing the supply base by identifying, selecting
and managing suppliers for strategic, long term partnerships is a “key ingredient to the
success of a supply chain”. In the SCM concept, a strategic, long-term cooperation in
partnership between the buyer and supplier should be reached to avoid waste of resources
within the logistics chain (Lasch and Janker, 2005). The main objective of supplier
selection process is to reduce purchase risk, maximise overall value to the purchaser and
develop closeness and long-term relationships between manufacturer and suppliers
(Li and Fun, 1997). However, selecting the right supplier is always a difficult task for
many purchasing managers (Liu and Hai, 2005). Supplier evaluation and selection
problem has been studied extensively and many research methodologies have been
proposed to tackle the problem. Supplier selection problem is always MCDM problem
and there has been comprehensive effort to develop decision making model for these
problems. The models which can integrate qualitative and quantitative criteria are best for
the practical application.
Supplier or vendor selection decisions are complicated by the fact that various criteria
must be considered in decision making process. These are the criteria/factors on which
the suppliers are evaluated. Supplier criteria are divided into quantitative and qualitative
attributes. A few of the classic and highly cited works (Busch, 1962; Dickson, 1966)
established that criteria like ‘quality’, ‘delivery schedule’ and ‘warranties’ are critical
factors for supplier selection amongst many others like ‘price’, ‘production capability’,
‘technical capability’, ‘management capability’, ‘vendor reputation’, ‘financial position’,
‘labour relations’, ‘post sales services’ and multiple other relationship specific attributes
like ‘past business records’ and ‘reciprocal arrangements’. Dickson (1966) in his seminal
460 N. Asthana and M. Gupta

work has found 23 important attributes used for evaluation and selection of suppliers.
Embarking from Dickson’s work, many articles have been published in later years
considering other different factors (Lambert et al., 1997; Weber et al., 1991; Dickson,
1966). A review of evaluating criteria has been provided by Cheraghi et al. (2004), and
concluded that
1 quality is top most evaluating criteria followed by delivery, price, and service
2 it is found that reliability, flexibility, consistency and long-term relationship as
significant new entrants of critical success factors for supplier selection.
Although these factors vary from product to product and organisation to organisation, the
objective of the supplier selection process is to reduce risk and maximise the total value
for the buyer/customer, and it involves considering a series of strategic decision making
variables (Kannan and Haq, 2007). Various key factors that exist for the performance
evaluation of suppliers are: leadership, trust, commitment, coordination, communication,
power, involvement, conflict resolution techniques and resources (Kim et al., 2010;
Giannakis, 2007). Evaluation and selection of suppliers based on the issues of cost,
quality and delivery are identified as the most important operational issues related to
managing sources of supply (Tummala and Schoenherr, 2011). Simpson et al. (2002)
carried out a survey to analyse evaluation processes for measuring the performance of
suppliers and reported that quality and process control are the most acceptable and widely
used evaluation criteria to assess the performance of suppliers.
Supplier selection methods are the models or approaches used to conduct the
selection process (Li and Fun, 1997). Various MCDM techniques have been proposed to
evaluate and select the suppliers from all the available suppliers. A review of MCDM
approaches/methods have been proposed by Agarwal et al. (2011), Boer et al. (2001),
Dey et al. (2010) and Khodadadzadeha and Sadjadib (2013). There are many methods
used in supplier selection, such as categorical methods (qualitative models), DEA, cluster
analysis, CBR systems, decision models like: linear weighting models, total cost of
ownership models, mathematical programming models, statistical models, and artificial
intelligence-based models (Boer et al., 2001). These approaches/models can be applied
individually to deal with supplier selection problem or can be integrated with some other
approaches to deal with supplier selection problem. Artificial intelligence-based models
such as ANN have been used extensively to deal with supplier selection problems
because it permits solutions to problems where multiple constraints must be satisfied
simultaneously, easily integrated with other methods, easily implemented to the software,
gives more precise results, possess the capability to generalise, can predict new outcomes
on past trends, exhibit mapping capabilities, can process information at high speed. ANN
is very useful in practical and dynamic applications, Lisboa et al. (1999) presented
application area of ANN, information about the neural network models and other such
methods against which they have been compared, are provided in the literature. Supplier
selection using ANN individually is represented by Jinlong et al. (1997), in this suppliers
are compared based on an index called selection coefficient. Creese et al. (2009)
presented a model based on integration of ANN and GA to select the supplier, here GA
was responsible for finding the weights of the neural network model. ANN was then used
to select the best supplier. Lau et al. (2006) used an integrated approach of ANN and GA
for supplier selection, ANN was responsible for benchmarking the potential suppliers
with respect to four evaluating factors and GA was deployed to determine the best
 Supplier selection using artificial neural network and genetic algorithm 461

combination of suppliers. A hybrid method using AHP and ANN for supplier selection
was provided by Ariffin et al. (2013), Kumar and Roy (2010) and Lakshmanpriya et al.
(2013) where AHP is used to determine the weights of criteria, and the ANN to select the
best supplier. Bayraktar and Celebi (2008) presented an integration of ANN and DEA
methods for evaluation of suppliers under incomplete information of evaluation criteria.
An analysis on various activation functions of neural network was presented by Jones
et al. (2013). Sigmoid activation function was found to give minimum error for the
number of epochs. Gupta and Sexton (2000) presented a comparison between GA and
back propagation (BP) for training ANN suggesting GA is superior than BP for training
of ANN.
Another artificial intelligence-based models/approach to solve supplier selection
problems are EA such as the GA is very widely used. GA is search and optimisation
algorithm that derive their computational mechanisms from natural selection and natural
adaptation. GA uses mechanisms inspired by biological evolution, such as reproduction,
selection, crossover, mutation. GA has been widely used because it can solve
multi-dimensional, non-differential, non-continuous, and even non-parametrical
problems, GA searches from one population of solutions to another, rather than from
individual to individual, GA only uses objective function information to guide themselves
through solution space and not derivatives, GA are easily transferred to existing
simulations and models. Arsovski et al. (2011) used GA to solve supplier selection
problem. Rungreunganaun and Woarawichai (2013) applied GA for inventory lot sizing
problem with supplier selection, the aim was to calculate the optimal inventory lot sizing
for each supplier and minimise total inventory cost, the problem was formulated as the
mixed integer programming and solved using GA. A multi objective supplier selection
problem has been formulated and solved by Basnet and Weintraub (2009), GA was used
to solve the bi-criteria problem, i.e., minimisation of cost and minimisation of the number
of suppliers. Kai et al. (2012) suggested a new heuristic GA method to solve supplier
selection problem. Sagar and Singh (2012) proposed a multi objective suppliers selection
model based on GA by considering incentive and penalty system.

3 Methodology

As already discussed in the above literature review that their exits number of criterion for
supplier selection and that they also vary from product to product and organisation to
organisation therefore the next step is the selection of supplier selection criteria. Since the
data was to be collected from the automobile organisation, therefore the most dominant
criteria in the particular sector are selected for supplier selection and are: quality, delay
time, unit cost, quantity, and service. The above selected criteria are in accordance with
the discussion held with the executives of the automobile sector.
The next step is the selection of the methodology for supplier selection. Among the
existing methodologies, GA and ANN are integrated together for the MCDM problem.
The reason for using the ANN model to solve supplier selection problem are: ANN
permits solutions to problems where multiple constraints must be satisfied
simultaneously, weights in the ANN can be modified with any EAs/other techniques very
easily, ANN possess the capability to generalise, can predict new outcomes on past
trends, ANN can be easily integrated with other methods, ANN exhibit mapping
capabilities. The reason for using GA in supplier selection process are: GA can solve
462 N. Asthana and M. Gupta

multi-dimensional, non-differential, non-continuous, and even non-parametrical

problems, GA searches from one population of solutions to another, rather than from
individual to individual, GA solves problems with multiple solutions, GA only uses
objective function information to guide themselves through solution space and not
derivatives, GA are easily transferred to existing simulations and models, GA can easily
solve the problem if number of parameters are very large.
Therefore the final decision making model in this study consists of the integration of
GA and ANN. The proposed model consists of two parts:
Part I applies GA to find the optimal value of function of quality, delay time, unit
cost, quantity, service. Linear objective functions of the above criteria of the form ∑WiXi
are formed. Weights ‘Wi’ are obtained by relative comparison of data of various
suppliers. Based on the industrial data and company policy bounds of variable ‘Xi’ are
also defined.
Part II uses the optimum value of the functions in the ANN model to determine the
supplier score. Best supplier is one which has minimum score. Figure 1 represents the
neural network decision making model which is formed to find out the supplier score
based on the input criteria’s.

Figure 1 Single layer feed forward neural network

WEIGHT, Wi
QUALITY
U1

DELAY TIME
U2

VENDOR
UNIT COST SCORES
U3 (OUTPUT)

QUANTITY
U4

SERVICE
U5

INPUT LAYER OUTPUT LAYER

The step by step procedure for calculation of supplier score is given below:
1 Calculation of weights of neural network for both input and output layer. The input
layer weights are calculated by pair wise comparison of the criteria’s. The output
layer weights (Wi) are calculated by relative comparisons of suppliers with respect to
particular criteria.
2 Formulation of the objective function, with constraints. An objective function of the
form
Minimise ∑ Wi Xi or Maximise ∑ Wi Xi
subject to W1 + W2 …… Wi = 1; L < X i < U

is formed. Here ‘Xi’ are the criteria’s and ‘L’, ‘U’ are its lower and upper bound.
 Supplier selection using artificial neural network and genetic algorithm 463

3 Optimisation of formulated objective function using GA. The optimisation tool box
of MATLAB is used to optimise the function, and find the optimum value of
particular criteria for each supplier.
4 Using the optimised value obtained from GA as an input to the INPUT layer for
calculation of the output of the INPUT layer.
5 Using the output value of the INPUT layer as an input for OUTPUT layer for
calculation of the output of the OUTPUT layer i.e. supplier score.
6 Validating the model with the manager/supervisor’s ranking of suppliers through
industrial visits and discussion with the executives of the organisation.

4 Calculations

Table 1 shows the data has been obtained from the automobile manufacturing company
dealing in manufacturing of both heavy duty and light duty vehicles.
Table 1 Supplier data

Supplier Quality Delay time Cost Quantity Service

S1 10 4 2190 400 G
S2 5 2 2800 500 G
S3 20 5 1800 200 VG
S4 10 3 2400 400 AV

Quality is in terms of ‘percentage defectives’, delay time in ‘days’, cost represented here
is per unit cost. G – good, VG – very good, AV – average.
Weights for input value can be found by pair wise comparison of criteria. In this
industry, quality was given preference over other criteria’s, followed by delay time, unit
cost, quantity, and service.
Let as assume that:
• quality is somewhat important than delay time – 3
• quality is more important than unit cost – 5
• quality is much more important than quantity – 7
• quality is very much more important than service – 9.

Table 2 Pair wise comparison of criteria’s

Quality Delay time Cost Quantity Service

Quality 1 3 5 7 9
Delay time 1/3 1 3 5 7
Cost 1/5 1/3 1 3 5
Quantity 1/7 1/5 1/3 1 3
Service 1/9 1/7 1/5 1/3 1
∑ 1.787 4.676 9.533 16.333 25
464 N. Asthana and M. Gupta

Table 3 Weights of criteria’s

Quality Delay time Cost Quantity Service Avg

Quality 0.5595 0.6415 0.5244 0.4285 0.3600 0.5028
Delay time 0.1865 0.2138 0.3146 0.3061 0.2800 0.2602
Cost 0.1119 0.0712 0.1048 0.1836 0.2000 0.1343
Quantity 0.0799 0.0427 0.0349 0.0612 0.1200 0.0677
Service 0.0621 0.0305 0.0209 0.0204 0.0400 0.0348

Based on the pairwise comparison of the criteria as shown in Table 2 and the subsequent
calculation of their weights as shown in Table 3, the value of CR is calculated.
Consistency index (C.I.) = 0.060652.
Random consistency index (R.I.) = 1.12.
Consistency ratio (C.R.) = 0.054153 < 0.1.

These obtained weights are input weights U1, U2, U3, U4, U5 for the input layer of the
neural model.
According to organisational policy:
• Maximum allowable percentage defective is 25%, so total scale is divided from
1% to 25%. Maximum allowable delay time is 8 days, so scale is divided from
0 to 8 days.
• For unit cost the difference between maximum and minimum cost is Rs 1,000 so
total scale is divided from 0–1,000.
• For quantity the difference between maximum and minimum is 300, so total scale is
divided between 0–300.
Table 4 shows the scale of quality, delay time, cost and quantity as per organisational
policy.
Table 4 Scales as per organisational policy

Quality (% def) Delay time (days) Cost Quantity Scale

0–5 0 0–100 0–60 1
6–10 1 101–200 61–90 2
11–12 2 201–300 91–120 3
13–14 3 301–400 121–150 4
15–16 4 401–500 151–180 5
17–18 5 501–600 181–210 6
19–20 6 601–700 211–240 7
21–22 7 701–800 241–270 8
23–25 8 801–1000 271–300 9

For service, Likert scale is used, very poor – 1, poor – 2, average – 3, good – 4,
very good – 5. If the difference between service of two suppliers is 0 then we scale as 1
similarly for 1 – 2, 2 – 3, 3 – 4, 4 – 5.
 Supplier selection using artificial neural network and genetic algorithm 465

Weights (Wi) of the output layer are obtained from the relative comparison of
suppliers for each criteria. Here for the evaluating criteria quality, we compare the
suppliers as shown in Table 5. On comparing (say) S1 and S3 we see that the difference
between the quality (% defective) is ‘10’ so from the scale we denoted it as ‘2’. Similarly
all others suppliers are compared.
Table 5 Relative matrix of suppliers with respect to quality

S1 S2 S3 S4
S1 1 1 2 1
S2 1 1 5 1
S3 1/2 1/5 1 1/2
S4 1 1 2 1
∑ 3.5 3.2 10 3.5

S1 S2 S3 S4 Avg
S1 0.285 0.3125 0.2 0.285 0.2706
S2 0.285 0.3125 0.5 0.285 0.3456
S3 0.143 0.0625 0.1 0.143 0.1121
S4 0.285 0.3125 0.2 0.285 0.2706
Note: Consistency ratio = 0.007675.
These are the weights obtained with respect to quality (% defective). Similarly the
relative weights can be obtained for all the criteria’s and these are represented in the final
weight matrix table as shown in Table 6.
Table 6 Weight matrix of suppliers with respect to all criteria’s

Quality Delay time Cost Quantity Service

S1 0.2706 0.1611 0.2430 0.1357 0.2270
S2 0.3456 0.4656 0.0425 0.0548 0.2270
S3 0.1121 0.096 0.5995 0.6737 0.1224
S4 0.2706 0.2772 0.1143 0.1357 0.4229

The value of CR for weight matrix of suppliers with respect to all criteria’s are given in
Table 7.
Table 7 CR for weight matrix of suppliers with respect to all criteria’s

Criteria C.R.
Quality 0.007675
Delay time 0.011507
Cost 0.07246
Quantity 0.039745
Service 0.003812
466 N. Asthana and M. Gupta

4.1 Calculations of GA
The industry will try to minimise % defective, delay time, cost and maximise quantity
and service, so the linear objective function of the form ∑WiXi for each criteria is formed
with lower and upper bound defined and then it is optimised using GA. Weights Wi are
calculated above. The lower and upper bound are defined by the industry. For solving the
equation formed using GA MATLAB software is used. The equations formed are:
• function of quality
Minimise f Quality = ∑ Wi Qi
= 0.2706Q1 + 0.3456Q2 + 0.1121Q3 + 0.2706Q4 .
0 < Q1 , Q 2 , Q3 , Q 4 < 25

• function of delay time

Minimise f DelayTime = ∑ Wi Di
= 0.1611D1 + 0.4656D2 + 0.096D3 + 0.2772D4 .
0 < D1 , D 2 , D3 , D 4 < 8.

• function of unit cost

Minimise f UnitCost = ∑ Wi Ci
= 0.2430C1 + 0.0425C2 + 0.5995C3 + 0.1143C4 .
1,500 < C1 , C2 , C3 , C4 < 3,000.

• function of quantity
Maximise f Quantity = ∑ Wi Qu i
= 0.1357Qu1 + 0.0548Qu 2 + 0.6737Qu 3 + 0.1357Qu 4 .
100 < Qu1 , Qu 2 , Qu 3 , Qu 4 < 1, 000.

• function of service
Maximise fService = ∑ Wi Si
= 0.2270S1 + 0.2270S2 + 0.1224S3 + 0.4229S4 .
1 < S1 , S2 , S3 , S4 < 5.

The above equations are optimised using GA toolbox in MATLAB software. The
algorithms parameter settings are as follows:
• population size = 20
• scaling function = rank
• selection = roulette wheel
• crossover fraction = 0.8
• mutation = adaptive feasible
• crossover function = scattered
 Supplier selection using artificial neural network and genetic algorithm 467

• hybrid function = none

• stopping criteria = 30 iteration
= 0.0001 (tolerance limit) which ever occur earlier.
Table 8 shows the optimum value of criteria obtained from GA and these function value
obtained are used as an input to the input layer of the ANN decision making model.
Table 8 Optimum value of criteria

Criteria’s Optimum value

Function of quality 0.017411
Function of delay time 0.044799
Function of cost 1497.6
Function of quantity 896.943
Function of service 4.97429

4.2 Calculations of ANN

Assumptions
1 Single layer feed forward neural network is used.

2 SIGMOIDAL function is selected.

Φ(I) = 1 (1 + e −αI ) ; α = slope parameter; I = ( ∑ Wi Xi + Bias ) .

It varies between 0 and 1. It gives minimum error in number of iteration.

3 Unsupervised learning has been selected.

4 A bias unit with input value 1 and weight 0.3 is used for both input and output layer.

Input layer calculation

Output of input layer Yi = [1 / (1 + e – α(ΣXiUi))].
Table 9 shows the calculations for the output of the input layer.
Table 9 Input layer calculations

Input Xi Weight Ui Bias Output of input layer (Yi)

0.017411 0.5028 0.3 0.5765
0.044799 0.2602 0.3 0.5772
1497.6 0.1343 0.3 1
896.943 0.0677 0.3 1
4.97429 0.0348 0.3 0.6161
468 N. Asthana and M. Gupta

Output layer calculation

Output of input layer Yo = [1 / (1 + e – α(ΣYiWi))].
Table 10 shows the calculations for the supplier score based on the input to the output
layer.
Table 10 Output layer calculations

Input to output layer 0.5765 0.5772 1 1 0.6161 Score

S1 0.2706 0.1611 0.2430 0.1357 0.2270 0.7441
S2 0.3456 0.4656 0.0425 0.0548 0.2270 0.7320
S3 0.1121 0.096 0.5995 0.6737 0.1224 0.8542
S4 0.2706 0.2772 0.1143 0.1357 0.4229 0.7552

Best supplier is one which is having minimum score.

• S2 is the best supplier.
• S2 > S1 > S4 > S3.

5 Model validation

The best judgment of the resulting solution can be obtained from the experts of the
system. The model results should be acceptable to the decision maker otherwise the
model may not be adequate for future use. The proposed model was validated with the
manager’s ranking of suppliers. In this industry the decision maker/the manager have
ranked the suppliers according to the number of times they have been given orders after
analysing all the evaluating criteria’s. The proposed model’s ranking of suppliers is in
total agreement to the manager’s ranking as shown in Table 11. This is due to the reason
that the supplier selection criteria were selected in accordance with the discussions with
the executives of the particular automobile industry, the data were from also collected
from the same organisation. Therefore, the proposed model in this paper is best suited to
the automobile sector of the Indian organisations and thus cannot be generalised for all
types of sectors. This result indicates that this model would be helpful to
managers/decision makers of the automobile sector to be consistent in their evaluation of
suppliers.
Table 11 Comparison of supplier’s score

Suppliers Final rank of suppliers (model result) Manager’s ranking of suppliers

S1 2 2
S2 1 1
S3 4 4
S4 3 3
 Supplier selection using artificial neural network and genetic algorithm 469

6 Discussion

In this paper, an integrated decision making model for supplier selection problem is
proposed using ANN and GA. The objective in this study is to integrate multi-attribute
decision making models that give grades to suppliers on a set of criteria with the
mathematical programming techniques that model the constraints and an objective
function to select the best supplier. Quality, delay time, unit cost, quantity, service are
used as the evaluating criteria. GA is used to find the optimum function value of the
criteria functions and then ANN is used to find the supplier score. The data used for the
evaluation of supplier is obtained from an automobile industry. Supplier score is the
output of the developed integrated model, based on the output score obtained the supplier
having the minimum score is the best supplier from the all available potential suppliers.
These suppliers are than ranked according to the score. The results obtained from the
integrated decision making model were validated with the decision maker/manager
ranking and are in agreement to the ranking as given by executives of the organisation.

7 Conclusions

This paper presents a study of the supplier selection problem. In this paper an attempt is
made to select the best and optimal supplier based on the evaluating criteria. The
integrated model developed is better in dealing with the complexity associated with the
supplier selection problem. The result of the study not only provide decision maker to
select the best supplier from the available potential suppliers but also rank the suppliers,
so that the decision maker could go for the alternatives.
The proposed model is best suited for the automobile sector and could not be
generalised since the criteria as well as data are collected from the automobile sector.
This counts to the major limitation of this study. As a research for future work this type
of study can be generalised considering some more criteria and feedback from the
executives of the other sector. Another major limitation of this work is that it deals with
only limited number of parameters for supplier selection. Some more parameters like
technical capability, financial position, production facilities and capacity, reliability,
performance history etc can be included in the study to give a better view of the
optimised results which will count to the scope of future research for the work.

Acknowledgements

The authors would like to thank the organisation and all the executives who provided us
the data and their all kinds of support in completing and validating the proposed
project/model. The authors also express their deeply thanks and gratitude to the
anonymous reviewers of the paper who have helped in the great deal to improve the
presentation of this paper.
470 N. Asthana and M. Gupta

References
Agarwal, P., Bag, M., Mishra, V., Sahai, M. and Singh, V. (2011) ‘A review of multi-criteria
decision making techniques for supplier evaluation and selection’, International Journal of
Industrial Engineering Computations, Vol. 2, No. 1, pp.801–810.
Ariffin, M.K.A., Hakim, N., Khaksar, W., Pah, P.S., Sulaiman, S. and Tang, S.H. (2013) ‘A hybrid
method using analytic hierarchical process and artificial neural network for supplier selection’,
International Journal of Innovation, Management and Technology, Vol. 4, No. 1, pp.109–111.
Arsovski, S., Arsovski, Z., Kalinic, Z., Milanovic, I., Petrovic, D.R. and Rankovic, V. (2011)
‘Multiobjective supplier selection using genetic algorithm: a comparison between weighted
sum and SPEA method’, International Journal for Quality Research, Vol. 5, No. 4,
pp.289–295.
Arunkumar, N., Karunamoorthy, L. and Makeshwaara, N. (2007) ‘An optimization technique for
vendor selection with quantity discounts using genetic algorithm’, International Journal of
Industrial Engineering, Vol. 3, No. 4, pp.1–13.
Azeem, A. and Paul, S.K. (2011) ‘An artificial neural network model for optimization of finished
goods inventory’, International Journal of Industrial Engineering Computations, Vol. 2,
No. 1, pp.431–438.
Basnet, C. and Weintraub, A. (2009) ‘A genetic algorithm for a bicriteria supplier selection
problem’, International Transaction in Operation Research, Vol. 16, No. 1, pp.173–187.
Bayraktar, D. and Celebi, D. (2008) ‘An integrated neural network and data envelopment analysis
for supplier evaluation under incomplete information’, Expert Systems with Applications,
Vol. 35, No. 1, pp.1698–1710.
Boer, D.L., Eva, L. and Pierangela, M. (2001) ‘A review of methods supporting supplier selection.’
European Journal of Purchasing & Supply Management, Vol. 7, No. 1, pp.75–89.
Busch, G. (1962) ‘New twist on supplier evaluation’, Journal of Purchasing, Vol. 55, No. 1,
pp.102–103.
Cheraghi, S.H., Dadashzadeh, M. and Subramanian, M. (2004) ‘Critical success factors for supplier
selection: an update’, Journal of Applied Business Research, Vol. 20, No. 2, pp.91–108.
Choy, K.L. and Lee, W.B. (2003) ‘A generic supplier management tool for outsourcing
manufacturing’, Supply Chain Management: An International Journal, Vol. 8, No. 2,
pp.140–154.
Creese, R.C., Golmohammadi, D., Kolassa, J. and Valian, H. (2009) ‘Supplier selection based on a
neural network model using genetic algorithm’, IEEE Transactions on Neural Networks,
Vol. 20, No. 9, pp.1504–1519.
Demirtas, E.A. and Ustun, O. (2008) ‘An integrated multiobjective decision making process for
supplier selection and order allocation’, Omega, Vol. 36, No. 1, pp.76–90.
Dey, P.K., Ho, W. and Xu, X. (2010) ‘Multi-criteria decision making approaches for supplier
evaluation and selection: a literature review’, European Journal of Operation Research,
Vol. 202, No. 1, pp.16–24.
Dickson, G.W. (1966) ‘An analysis of vendor selection systems and decisions’, Journal of
Purchasing, Vol. 2, No. 1, pp.5–17.
Giannakis, M. (2007) ‘Performance measurement of supplier relationships’, Supply Chain
Management: An International Journal, Vol. 12, No. 6, pp.400–411.
Gulen, K.G. (2007) ‘Supplier selection and outsourcing strategies in supply chain management’,
Journal of Aeronautics and Space Technologies, Vol. 3, No. 2, pp.1–6.
Gupta, M. and Narain, R. (2012) ‘A survey on supplier relationship in e-procurement in Indian
organisations’, International Journal of Logistics Systems and Management, Vol. 12, No. 1,
pp.89–121.
Gupta, N.D.J. and Sexton, R.S. (2000) ‘Comparative evaluation of genetic algorithm and
backpropagation for training neural metwork’, International Journal of Information Sciences,
Vol. 129, No. 1, pp.45–59.
 Supplier selection using artificial neural network and genetic algorithm 471

Jinlong, Z., Siying, W. and Zhicheng, L. (1997) ‘A supplier selecting system using neural
network’, IEEE International Conference on Intelligent Processing Systems, Beijing, China,
pp.468–471.
Jones, S.A., Sibi, P. and Siddarth, P. (2013) ‘Analysis of different activation functions using back
propagation neural networks’, Journal of Theoretical and Applied Information Technology,
Vol. 47, No. 3, pp.1264–1268.
Kai, L.M., Lixin, C. and Wei, S. (2012) ‘An improved genetic approach to optimal supplier
selection and order allocation with customer flexibility for multi-product manufacturing’,
Industrial Engineering & Management Systems, Vol. 11, No. 2, pp.155–164.
Kannan, G. and Haq, A.N. (2007) ‘Analysis of interaction of criteria and sub-criteria for the
selection of supplier in the built-in-order supply chain environment’, International Journal of
Production Research, Vol. 45, No. 17, pp.3831–3852.
Khodadadzadeha, T. and Sadjadib, S.J. (2013) ‘A state-of-art review on supplier selection
problem’, Decision Science Letters, Vol. 2, No. 1, pp.59–70.
Kim, D.Y., Kumar, V. and Kumar, U. (2010) ‘Performance assessment framework for supply
chain partnership’, Supply Chain Management: An International Journal, Vol. 15, No. 3,
pp.187–195.
Kouvelis, P., Rosenblatt, M.J. and Munson, C.L. (2004) ‘A mathematical programming model for
global plant location problems: analysis and insights’, IIE Transactions, Vol. 36, No. 2,
pp.127–144.
Kumar, J. and Roy, N. (2010) ‘A hybrid method for vendor selection using neural network’,
International Journal of Computer Applications, Vol. 11, No. 12, pp.35–40.
Lakshmanpriya, C., Lavanpriya, C. and Sangeetha, N. (2013) ‘Vendor selection in manufacturing
industry using AHP and ANN’, The SIJ Transactions on Industrial, Financial & Business
Management (IFBM), Vol. 1, No. 1, pp.29–34.
Lambert, D.M., Adams, J.R. and Emmelhainz, M.A. (1997) ‘Supplier selection in the healthcare
industry: a comparison of importance and performance’, International Journal of Purchasing
and Materials Management, Vol. 33, No. 1, pp.16–23.
Lasch, R. and Janker, C.G. (2005) ‘Supplier selection and controlling using multivariate analysis’,
International Journal of Physical Distribution & Logistics Management, Vol. 35, No. 6,
pp.409–425.
Lau, H.C.W., Lee, C.K.M., Ho, G.T.S., Pun, K.F. and Choy, K.L. (2006) ‘A performance
benchmarking system to support supplier selection’, International Journal of Business
Performance Management, Vol. 8, Nos. 2/3, pp.132–151.
Levary, R.R. (2007) ‘Ranking foreign suppliers based on supply risk’, Supply Chain Management:
An International Journal, Vol. 12, No. 6, pp.392–394.
Li, C.C. and Fun, Y.P. (1997) ‘A new measure for supplier performance evaluation’, IIE
Transactions, Vol. 29, No. 1, pp.753–758.
Lisboa, P.J.G., Vaughan, J. and Vellido, A. (1999) ‘Neural networks in business: a survey of
applications (1992–1998)’, Expert Systems with Applications, Vol. 17, No. 1, pp.51–70.
Liu, F.H.F. and Hai, H.L. (2005) ‘The voting analytic hierarchy process method for selecting
supplier’, International Journal of Production Economics, Vol. 97, No. 3, pp.308–317.
Narasimhan, R., Jayaram, J. and Carter, J.R. (2001) ‘An empirical examination of the underlying
dimensions of purchasing competence’, Production and Operations Management, Vol. 10,
No. 1, pp.1–15.
Ng, W.L. (2008) ‘An efficient and simple model for multiple criteria supplier selection problem’,
European Journal for Operation Research, Vol. 186, No. 3, pp.1059–1067.
Pal, P. and Kumar, B. (2008) ‘16T: toward a dynamic vendor evaluation model in integrated SCM
processes’, Supply Chain Management: An International Journal, Vol. 13, No. 6, pp.391–397.
Rungreunganaun, V. and Woarawichai, C. (2013) ‘Applying genetic algorithm for inventory lot
sizing problem with supplier selection under storage space’, World Academy of Science,
Engineering and Technology, Vol. 74, No. 1, pp.303–308.
 472 N. Asthana and M. Gupta

Sagar, M.K. and Singh, A.P. (2012) ‘An optimization technique for selection and rating of
suppliers with incentive & penalty system using genetic algorithm’, International Journal of
Research in Engineering & Applied Sciences, Vol. 2, No. 7, pp.72–85.
Simpson, P.M., Siguaw, J.A. and White, S.C. (2002) ‘Measuring the performance of suppliers: an
analysis of evaluation processes’, Journal of Supply Chain Management, Vol. 38, No. 1,
pp.29–41.
Talluri, S. and Narasimhan, R. (2004) ‘A methodology for strategic sourcing’, European Journal of
Operational Research, Vol. 154, No. 1, pp.236–250.
Tangpong, C., Hung, K.T. and Ro, Y.K. (2010) ‘The interaction effect of relational norms and
agent cooperativeness on opportunism in buyer-supplier relationships’, Journal of Operations
Management, Vol. 28, No. 5, pp.398–414.
Tummala, V.M.R. and Schoenherr, T. (2011) ‘An implementation decision framework for supply
chain management: a case study’, International Journal of Logistics Systems and
Management, Vol. 8, No. 2, pp.198–213.
Weber, C.A., Current, J.R. and Benton, W.C. (1991) ‘Vendor selection criteria and methods’,
European Journal of Operational Research, Vol. 50, No. 1, pp.2–18.

View publication stats