AN INTELLIGENT DECISION SUPPORT SYSTEM FOR

SUPPLIER SELECTION

R. J. Kuo, L. Y. Lee

Department of Industrial Engineering and Management, National Taipei University of Technology

No. 1, Section 3, Chung-Hsiao East Road, Taipei, Taiwan

Tung-Lai Hu

Department of Business Management, National Taipei Universit of Technology

No. 1, Section 3, Chung-Hsiao East Road, Taipei, Taiwan

Keywords: Decision support system, Supplier selection, Fuzzy DEA, Fuzzy AHP.

Abstract: This study intends to develop an intelligent decision support system which integrates both fuzzy analytical

hierarchy process (AHP) method and fuzzy data development analysis (DEA) for assisting organizations to

make the supplier selection decision. A case study on an internationally well-known auto lighting OEM

company shows that the proposed method is very well suitable for practical applications.

1 INTRODUCTION

Supply chain management consists of several

connected logistics systems, which integrate the

product and service moving into a system, and

creates a continuous and seamless linking. Also, all

the actions from raw materials to end customers for

merchandises are fully coordinated. Due to such

coordination, all the members inside the supply

chain will be affected by other chain members either

directly or indirectly. Therefore, it is very important

to select suitable suppliers to overcome these

problems. Regarding the supplier selection, some

indicators, like production capacity, financial

capability, quality, etc. should be put into account.

Otherwise, supplier problem may become

organization’s crisis.

This study intends to present a novel

performance evaluation method which integrates

both fuzzy AHP method and fuzzy DEA for

organizations to make the supplier selection

decision. Though DEA has been applied in the area

of performance evaluation for many decades, it

posses its own limitations. They include a method to

determine the weight constraints and integrity of the

evaluation data. Definitely, the number of evaluation

samples should also be two times of the sum of input

and output numbers. In this study, we try to

overcome these limitations. First, using fuzzy AHP

method can find the indicators’ weights. Then

cut set and extension principle of fuzzy set theory

simplifies the fuzzy DEA as a pair of traditional

DEA model with

-cut level. Finally, fuzzy

ranking using maximizing and minimizing set

method is able to rank the evaluation samples.

A case study on an internationally well-known

auto lighting manufacturer showed that the proposed

method was more suitable for the practical

applications after comparing with the traditional

fuzzy DEA method. The case company is able to use

the computational results to adjust her suppliers’

inputs in order to obtain more promising

performance.

2 BACKGROUND

This section will briefly present the general

background of supplier evaluation, fuzzy AHP and

fuzzy DEA.

2.1 Supplier Evaluation

Suppliers are the vendors who provide raw materials,

components or service that an organization itself

241

J. Kuo R., Y. Lee L. and Hu T. (2008).

AN INTELLIGENT DECISION SUPPORT SYSTEM FOR SUPPLIER SELECTION.

In Proceedings of the Tenth International Conference on Enterprise Information Systems - AIDSS, pages 241-248

DOI: 10.5220/0001694202410248

 SciTePress

cannot offer. The selection of right suppliers is the

first step of supply chain evaluation. In the current

manufacturing environment for supply chain,

suppliers are a vital part for an organization and a

right supplier can furnish the company with quality

products of required quantity at reasonable prices

before the predetermined delivery schedule to

sharpen company competitiveness and quicken

company response to market and customer demands.

There have already been a number of methods

proposed to evaluate the suppliers. They are Monte

Carlo Simulation (Thompson, 1990), mathematical

programming (Weber and Current, 1993), AHP

(Mohanty and Deshmukh, 1993), DEA (Narasimhan

et al., 2001), integrated AHP and DEA (Liu et al.,

2005), and neural network (Kuo et al., 2008).

The attributes of indicators have great impact on

the evaluation results. Among the 23 indicators

proposed by Dickson (1966), quality, delivery

deadline and previous performance are given

primary importance in the 1960’s manufacturing.

After that, a number of researches proposed different

indicators for assessing suppliers. They can be found

in (Lehmann and O’Shaughnessy, 1982, Wilson,

1994, Weber et al., 1991, Smytka and Clemens,

1993, Swift, 1995, Choi and Hartley, 1996, Goffin et

al., 1997, Narasimhan et al., 2001, Quayle, 2002,

Schmitz and Platts, 2004, and Wang et al., 2004).

After summarizing the evaluation indicators

proposed in the earlier literatures, it is revealed that

the supplier company itself should also be taken into

consideration in addition to the product and delivery

quality. These include the supplier’s organizational

structure, management and financial status.

2.2 Data Envelopment Analysis

DEA, which was developed by Charnes, Cooper and

Rhodes (Banker et al., 1984), is a mathematical

programming technique for measuring the relative

performance of decision making units (DMUs) on

the basis of the observed operation practice in a

sample of comparable DMUs. It has typically been

applied to analyze the relative production efficiency

of DMUs in a setting of multiple incommensurate

input and output variables.

The standard DEA model is selected as the

reference unit. For each DMUs, the composite unit

that consumes the lowest possible fraction of that

DMU’s current input levels to produce at least that

DMU’s current output levels. More formally, the

reference units are identified simultaneously by

solving the linear programming problem. By

comparing n units with m inputs denoted by

m)1,...,(ix

and outputs denoted

s)1,...,(ry

, the efficiency measure for DMU

k follows the linear programming problem as

m1,...,i s;1,...,r ;u,v

n 1,...,j 1,

s.t.

hmax

iji

rjr

iki

rkr

==≥

=≤

∑

(1)

h = the relative efficiency of the kth DMU

y The rth output value of the kth DMU

u The virtual multiplier of the rth output value

x The ith input value of the kth DMU

v The virtual multiplier of the ith input value

2.3 Fuzzy DEA

Sengupta (1992) first incorporated the fuzzy theory

into DEA in 1992 and studied the performance

evaluation by using randomly found observation

values. Such a random DEA theory was further

examined by Cooper et al. (1996 and 1998) who

focused more on theoretical exploration. In 2000,

Kao and Liu proposed a supplier selection method

with indefinite data. By using the

-cut and the

extension principle of the fuzzy theory, they

simplified the Fuzzy DEA method into a pair of

traditional DEA analysis modes with

horizontal

parameters. The efficiency value acquired in this

study was fuzzy rather than definite as found

traditionally. Thus, it was hard to rank the

alternatives based on the efficiency values solely.

Kao and Liu proposed to use two fuzzy values

sequencing methods, i.e., “area estimation” and “the

maximum and minimum sets,” to rank the efficiency

values obtained by using different

-cut methods.

They revealed that “area estimation” is more

effective and easier in ranking the fuzzy values when

the exact form of the membership function remains

unknown. When the

-cut approaches infinity, the

method is theoretically faultless. While in the

practical application, it is critical to select a right

cut. Smaller

-cut values increase the effectiveness

of the method and adequately large

-cut

guarantees the correctness of the ranking. Therefore

an appropriate

-cut value is vital for the right

sequencing of the results.

ICEIS 2008 - International Conference on Enterprise Information Systems

242

3 METHODOLOGY

The proposed performance evaluation method

consisted of three components: determination of

indicators, fuzzy AHP and fuzzy DEA. Each part is

discussed in the following subsections. Improvement

analysis is made on some suppliers and comparison

is made against the results of the fuzzy DEA.

3.1 Determination of Supplier

Evaluation Indicators

For the purpose of obtaining more practical and

objective results, besides including the indicators in

the literatures, the case company’s opinions have to

be considered for determining the evaluation

indicators.

3.2 Determination of Indicator Weights

After determining the evaluation factors, fuzzy AHP

has been used for determining the indicator weights.

The decision makers fill in an interval number in

terms of the significance level and compare the

indicators in pairs. A fuzzy number has been added

and subtracted to ensure the goal of the results. The

decision makers can rank the significance levels as

1~2, 2~4, 4~6, 6~8, and 8~9, the fuzzy positive

reciprocal matrix. Based on the fuzzy positive

reciprocal matrix, Lambda-Max method proposed by

Csutora and Buckley (2001) is used to calculate the

fuzzy weight in the fuzzy AHP. To ensure that the

weights obtained are fuzzy, we use the adjustment

coefficient to get the upper and the lower limit

values of the weight for each dimension.

A consistency test shall be carried out to find

the Consistence Index (C.I.) to ensure the conformity

of the calculation results. In the positive reciprocal

matrix, slight changes of

result in minor

fluctuation of

max

. So the disparity between

max

and n is an indicator of the conformity. The

Consistency Ratio (C.R.) is defined as Equations (2)

and (3).

RC =

(2)

1-n

C.I.

max

(3)

The Random Indices (R.I) are random indicators.

When C.R.= 0, the prior and the later judgments are

consistent; the larger the C.R is, the larger the

disparity that exists. According to Saaty (1977), the

error when

0.1C.R.

≤

is acceptable.

3.3 Evaluation of Suppliers with DEA

The weights acquired in the first stage by using

fuzzy AHP are used here. The standardized data are

input for the calculation of fuzzy performance

indicators and supplier selection.

3.3.1 Determination of Suppliers under

Evaluation

The evaluation data on the previous suppliers of the

company can be used to select suppliers of new

components or in annual appraisal of all the

suppliers.

3.3.2 Definition of Input and Output

Indicators

The input items should cover the supplier’s

executive force, capability, quality system, flexibility

and relationship with other suppliers, based on which

the indicators are further expanded. The supplier

productivity and operation efficiency are the major

aspects of the output. In Fuzzy DEA, based on the

objective function of DEA, max

∑

, the larger the output using the

less input, the better the performance is. It is

necessary to transform the output indicators into

fractional numbers as large as possible and the input

indicators as small as possible.

3.3.3 Homogenized Data

Liu et al. (2005) addressed that the original DEA

mode sets no limit to the weight range, so the

original unit of the variables can be used, though the

data shall be homogenized into values within the

same value range so that the weight range of

evaluation indicators is meaningful. Each indicator

of different suppliers is differentiated as 1 at the

highest and other indicators vary at equal proportion.

3.3.4 Fuzzy DEA

Although DEA is an effective method for efficiency

evaluation, it fails to work out fuzzy values of

incomplete data or data of large disparity. Kao and

Liu (2000) developed a mode for such data by using

the

-cut and the extension principle of the fuzzy

AN INTELLIGENT DECISION SUPPORT SYSTEM FOR SUPPLIER SELECTION

243

theory to simplify the Fuzzy DEA mode into a

traditional DEA mode with

horizontal

parameters. At a given

, the upper and the lower

limit value of efficiency can be found and the

membership functions of the efficiency values can

be constructed by using the efficiency values at

different

levels. This method is similar to the

traditional DEA analysis mode in the calculation of

efficiency improvement, technological efficiency

and scale efficiency.

By assuming that

and

are fuzzy data,

they can be expressed as the membership functions

and

by using the fuzzy set theory. As for

non-fuzzy, or crisp, data, the membership function is

a degenerate one and the domain is limited to one

value. Based on CCR model, it can be expressed as:

m,1,i s,1,r v,u

n ,1,j , 0X

v-Y

s.t.

MaxE

ijirj

iki

rkr

==≥

=≤

∑∑

∑

(4)

3.3.5 Combining Fuzzy AHP with Fuzzy

DEA

Thereafter, we can put the range of indicators

obtained in Section 3.2 into fuzzy DEA model.

4 MODEL EVALUATION

RESULTS

This section will apply the proposed method,

integration of fuzzy AHP and fuzzy DEA methods,

for studying the company according to the

procedures presented in Section three. The company

in this study is an internationally well-known auto

lighting system OEM Company. Since the company

under investigation has many outsourcing

components, according to the purchasing manager’s

suggestion, this study selects speculum as the

studying object since it plays a very important role in

the auto lighting system design. Thus, the speculum

vendors constitute the evaluation population.

4.1 Indicator Determination

After referring to company’s current indicators,

literatures and company’s managers, this study

decided to formulate five main dimensions (the first

level of AHP) including supplier’s operation

capability, supplier’s capability, quality system,

flexibility, and supplier relationship. From these five

dimensions, we can extend a total of eleven

evaluation indicators (second level of AHP) as

illustrated in Table 1. The AHP structure is for input

items, while the output items are also shown in the

lower part of Table 1. It indicates that the company

emphasizes more on the production and operational

efficiencies, respectively. Production efficiency can

be divided into production line efficiency and

production employee’s efficiency, while supplier

asset operational efficiency and business volume are

the two major extensions for operational efficiency.

Basically, it is supposed that inputs and outputs are

closely related. The scoring of indicators is provided

by the company and the statistical data is collected

from 2003 to 2005.

4.2 Fuzzy Weight Determination of

Evaluation Indicators

There are a total of twelve respondents. Since each

person has different perception on the level of

importance, the questionnaire let the respondents

directly determine their own range values for each

importance level. Then the comparison matrix can be

formulated for calculation and to make a consistent

test. After summarization, the final results of fuzzy

weights are depicted in Table 1.

4.3 Data Normalization

Since there is no constraint for the weights’ ranges

of input and output variables, the original DEA

model can keep the original units for the variables.

However, due to the range of weight, it is necessary

to normalize the data and make them in the same

value range. This can be helpful to provide the

evaluation indicator weight range. Data

normalization is conducted so as to let indicators

have the largest value as one. For the selected

product item, the case company only has ten

suppliers totally.

4.4 Performance Evaluation Analysis

This study integrates both fuzzy AHP and fuzzy

DEA methods to evaluate speculum suppliers’

ICEIS 2008 - International Conference on Enterprise Information Systems

244

performance. In the above, we can apply fuzzy AHP

to get the range of weights for output indicators.

After putting them into fuzzy DEA model, we can

analyze the performance evaluation. From the

calculation of LINGO based on the concept of

cut, Table 2 lists the fuzzy performance value of

each supplier under the

-cut levels as 0, 0.1, 0.2,

0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, and 1.0.

Table 2 reveals that only supplier D has the

highest performance value. The other nine suppliers’

performance values are all smaller than 1. Therefore,

DEA’s dual model can help us solve the problem of

how to adjust the input and output indicators in order

to improve the performance. First, rank every

supplier’s fuzzy performances using maximizing and

minimizing set method. Then, get the total utility

value

)(iU

for every fuzzy efficiency value

as illustrated in Figure 1. It indicates that

JAEIGFHBCD

>>>>>>>>>

and supplier D is the best supplier.

4.5 Improving Efficiency Analysis

Above section has shown that supplier D is the best

supplier after sorting all the suppliers’ fuzzy

performance. Here, we take

=0 level, for instance,

in order to provide the possible range for improving

the target value. This is for those suppliers with low

performance. Basically, the difference of

performance scores is the largest as

is equal to 0

from examining each supplier’s fuzzy performance

under eleven

levels. In other words, when

analyzing and improving unknown or missing data,

it is feasible to consider improving and analyzing

both the relatively high-test and lowest efficiency

values. Through this procedure, we can know the

required improvement range for the supplier with

missing or unknown data. The adjusting equations

for inputs are illustrated in Equations (5) and (6).

)(s-)(x*)(x

(5)

)(s-)(x*)(x

(6)

Since the computation of fuzzy DEA with

indicators’ weights uses the normalized data for

analysis and the analysis improvement applies the

original scores for computation, it is necessary to

multiply

s and

with the largest value of all

the indicators during normalization while adjusting

input and output items. Table 3 presents the

improved goals under the levels of

=0.

4.6 Comparison with the Original

Fuzzy DEA

In order to prove the proposed method’s feasibility,

original DEA is adopted for the purpose of

comparison. Its model is identical to the current

study. The two major differences are that there is no

weight constraint for each indicator and it is not

necessary to normalize the data. The total utility

values of fuzzy performance using the original fuzzy

DEA are all 1s. It reveals that in the original fuzzy

DEA is not able to discriminate these suppliers,

since the number of DMUs is not more than two

times of the indicator number. After integrating

fuzzy AHP into the fuzzy DEA model, the proposed

model is able to discriminate the efficiency values.

In the original fuzzy DEA model, since the number

of suppliers is too few, there is no meaning for the

evaluation results. Thus, there should be at least

thirty suppliers if the original fuzzy DEA is applied.

Since the case company only has ten suppliers, this

is the reason for all the efficiency values being 1s.

From the practical view point, no matter in the

traditional or high-tech industries, it is very common

that there won’t be too many suppliers for a certain

component. Some company may only have two to

three suppliers for one component. The reason is that

the company prefers spending more time to maintain

each other’s relationship on a few suppliers instead

of having a lot of suppliers. Thus, the current

proposed method is more suitable for industrial

practical situations. Our results are also very

consistent to company’s own evaluation results after

discussing with the senior managers.

5 CONCLUSIONS

In this study, fuzzy DEA is employed for supplier

selection and evaluation and AHP with fuzzy values

are introduced to define the weight range of

indicators weight. In this way, significant indicators

are used in the supplier performance evaluation. The

two stages of fuzzy AHP and fuzzy DEA single out

the supplier to the best benefit of the company

effectively. This method is applicable to all

industries and is quite easy and simple if the

evaluation indicators and weights are available. The

calculation process is not so complicated. After one

supplier is chosen, the other suppliers with poor

performance can find improvement schemes.

Comparison is made against the fuzzy DEA without

including the evaluation indicators to expose the

AN INTELLIGENT DECISION SUPPORT SYSTEM FOR SUPPLIER SELECTION

245

impact of the evaluation indicators identified on the

supplier selection.

A case study on an internationally well-known

auto lighting system OEM Company showed that the

proposed method really has the above advantages.

Through the results provided by the proposed

method, the present company can make some

adjustments for her suppliers in order to obtain more

attractive outcomes.

REFERENCES

Banker, R.D., Charnes, A., and Cooper, W.W., 1984.

Some models for estimating technical and scale

inefficiencies in data envelopment analysis.

Management Science, 30(9), pp.1078-1092.

Choi, T.Y., Hartley, J.L., 1996. An explortation of supplier

selection practices across the supply chain. Journal of

Operations Management, 1, pp.333-343.

Cooper, W. W., Huang, Z., Lelas, V., Li, S. X. and Olesen,

O. B., 1998. Chance Constraint Programming

Formulations for Stochastic Characterizations of

Efficiency and Dominance in DEA. Journal of

Productivity Analysis, 9, pp.53-79.

Cooper, W. W., Huang, Z. and Li, S. X., 1996. Satisfying

DEA Models under Chance Constraints. Annals of

Operations Research, 66, pp.279-295.

Csutora R. and Buckley, J. J., 2001. Fuzzy hierarchical

analysis: The Lambda-Max method. Fuzzy Sets and

Systems, 120, pp.181-195.

Dickson, G.W., 1996. An analysis of vender selection

systems and decisions. Journal of Purchasing, 2(1),

pp.5-17.

Goffin, K., Szwejczewski, M. and New, C., 1997.

Managing suppliers: when fewer can mean more.

International Journal of Physical Distribution and

Logistics Management, 27(7), pp.422-436.

Kao, C. and Liu, S.-T., 2000. Fuzzy efficiency measures in

data envelopment analysis. Fuzzy Sets and Systems,

113, pp. 427-437.

Kuo, R.J., Hong, S.M., Lin, Y., and Huang, Y.C., 2008.

Continuous genetic algorithm based fuzzy neural

network for learning fuzzy if-then rules,” Accepted by

Journal of Neuralcomputing.

Lehmann, D. R. and O’Shaughnessy, J., 1982. Decision

criteria used in buying different categories of products.

Journal of Purchasing and Materials Management,

pp.9-14.

Liu C.-M., Hsu H.-S., Wang S.-T. and Lee H.-K., 2005. A

performance evaluation model based on AHP and

DEA. Journal of the Chinese Institute of Industrial

Engineers, 22(3), pp. 243-251.

Mohanty, R. P. and Deshmukh, S. G., 1993. Using of

analytic hierarchic process for evaluating sources of

supply. International Journal of Physical Distribution

& Logistics Management, 23(3), pp.22-28.

Narasimhan, R., Talluri, S and Mendez, D., Summer 2001.

Supplier evaluation and rationalization via data

envelopment analysis: an empirical examination. The

Journal of Supply Chain Management.

Quayle, M., 2002. Puchasing in small firms. Eurpean

Journal of Purchaing and supply Management, 8,

pp.151-159.

Saaty, T.L., 1977. A scaling method for priorities in

hierarchical structures. Journal of Mathematical

Psychology, 15, pp. 234-281.

Schmitz, J., Platts, K.W., 2004. Supplier logistics

performance measurement: indication from a study in

the automotive industry. International Journal of

Production Economics, 89, pp.231–243.

Sengupta Jati K., 1992. A fuzzy systems approach in data

envelopment analysis. Computers Mathematic

Application, 24(8/9), pp.259-266.

Swift, C. O., 1995. Performance for single sourcing and

supplier selection criteria. Journal of Business

Research, 32(2), pp.105-111.

Smytka, D.L., Clemens, M.W., 1993. Total cost supplier

selection model: a case study. International Journal of

Purchasing and Materials Management, 29(1), pp.42-

49.

Thompson, K.N., winter 1990. Vendor profile analysis.

International Journal of Purchasing and Materials

Management, pp.11-18.

Wang, G., Huang, S.H., Dismuke, J.P., 2004. Product-

driven supply chain selection using integrated multi-

criteria decision-making methodology. International

Journal of Production Economics, 91, pp.1-15.

Weber, C. A., Current, J. R. and Benton, W. C., 1991.

Vendor selection criteria and method. European

Journal of Operational Research, 50(1), pp2-18.

Weber, C. A. and Current, J. R., 1993. A multiobjective

approach to vendor selection. European Journal of

Operational Research, 68(2), pp.173-176.

Wilson, E. J., 1994. The relative importance of supplier

selection criteria：a review and update. International

Journal of Purchasing and Materials Management,

30(3), pp.35-41.

ICEIS 2008 - International Conference on Enterprise Information Systems

246

Table 1: The average weighted range for each evaluation indicator.

Input Indicators Range of Weights

Delivery schedule (

V )

0.381V0.264

≤≤

Implementation capability

Cost analysis (

V )

0.192V0.136

≤≤

R&D capability (

V )

0.124V0.073

≤≤

Manufacturing

capability

Manufacturing process capability (

V )

0.245V0.186

≤≤

Quality management system (

V )

0.173V0.104

≤≤

Manufacturing process inspection system (

V )

0.098V0.045

≤≤

Quality system

Outbound quality (

V )

0.380V0.253

≤≤

Emergency order processing capability (

V )

0.220V0.143

≤≤

Flexibility

Response speed of exceptional process (

V )

0.364V0.2129

≤≤

Supplier’s financial capability (

V )

0.129V0.076

≤≤

Supplier relationship

Supplier’s coordination (

)

0.2540.150

≤≤ V

Output indicators

Production line efficiency (

U )

0.602U0.436

≤≤

Production efficiency

Employee’s production efficiency (

U )

0.227U0.168

≤≤

Business volumn ratio (

U )

0.202U0.134

≤≤

Operation efficiency

Supplier’s asset operational efficiency (

U )

0.046U0.030

≤≤

Table 2: The fuzzy performance indicator value of each supplier under 11 different

-cut levels.

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

L 0.7510 0.7530 0.7549 0.7569 0.7588 0.7608 0.7628 0.7648 0.7668 0.7689

0.77

U 0.7981 0.7939 0.7898 0.7872 0.7848 0.7825 0.7801 0.7778 0.7755 0.7732

0.77

L 0.9679 0.9686 0.9693 0.9701 0.9768 0.9716 0.9722 0.9730 0.9737 0.9745

0.97

U 0.9802 0.9797 0.9792 0.9787 0.9782 0.9777 0.9772 0.9767 0.9762 0.9758

0.97

L 0.9975 0.9975 0.9975 0.9975 0.9975 0.9975 0.9975 0.9975 0.9975 0.9975

0.99

U 0.9975 0.9975 0.9975 0.9975 0.9975 0.9975 0.9975 0.9975 0.9975 0.9975

0.99

L 1 1 1 1 1 1 1 1 1 1 1

U 1 1 1 1 1 1 1 1 1 1 1

L 0.0001 0.8110 0.8110 0.8110 0.8110 0.8110 0.8110 0.8110 0.8110 0.8110

0.81

U 0.8110 0.8110 0.8110 0.8110 0.8110 0.8110 0.8110 0.8110 0.8110 0.8110

0.81

L 0.8110 0.9155 0.9212 0.9245 0.9274 0.9306 0.9338 0.9368 0.9400 0.9432

0.94

U 0.9155 0.9797 0.9759 0.9721 0.9683 0.9646 0.9609 0.9572 0.9536 0.9500

0.94

L 0.9836 0.8529 0.8584 0.8641 0.8698 0.8757 0.8816 0.8876 0.8936 0.8998

0.90

U 0.8743 0.9729 0.9673 0.9549 0.9458 0.9383 0.9315 0.9250 0.9156 0.9122

0.90

AN INTELLIGENT DECISION SUPPORT SYSTEM FOR SUPPLIER SELECTION

247

Table 2: The fuzzy performance indicator value of each supplier under 11 different

-cut levels. (cont.)

L 0.9823 0.9694 0.9694 0.9694 0.9694 0.9694 0.9694 0.9694 0.9694 0.9694

0.96

U 0.9694 0.9694 0.9694 0.9694 0.9694 0.9694 0.9694 0.9694 0.9694 0.9694

0.96

L 0.9694 0.7245 0.7245 0.7245 0.7245 0.7245 0.7245 0.7245 0.7245 0.7245

0.72

U 0.7245 0.7245 0.7245 0.7245 0.7245 0.7245 0.7245 0.7245 0.7245 0.7245

0.72

L 0.7245 0.6416 0.6463 0.6510 0.6559 0.6608 0.6658 0.6708 0.6760 0.6811

0.68

U 0.6369 0.7490 0.7410 0.7332 0.7255 0.7180 0.7106 0.7038 0.6949 0.6922

0.68

Table 3: The goal range after improvement of input indicators for supplier A under

=0 level.

Supplier Number

Original total score 10 1 1 50 20 20

Original score 4 0.95 0.89 36 14 14

(L) 0.751 0.751 0.751 0.751 0.751 0.751

(U) 0.798 0.798 0.798 0.798 0.798 0.798

−

(L) 0*7 0*1 0*0.3 0*15.5 0*6 0*7

−

(U) 0*7 0*1 0*0.3 0*15.5 0*6 0*7

Goal after improvement (L) 4.51 0.71 0.08 10.51 4.51 4.51

Goal after improvement(U) 4.79 0.76 0.09 11.17 4.79 4.79

Goal after actual improvement (

) 5.21 0.76 0.91 38.83 15.21 15.21

Goal after actual improvement (

) 5.49 0.71 0.92 39.49 15.49 15.49

Supplier Number

Original total score 1 1 10 5 20

Original score 0.72 0.75 5 3 [10.5,15.5]

(L) 0.751 0.751 0.751 0.751 0.751

(U) 0.798 0.798 0.798 0.798 0.798

−

(L) 0*0.4 0*0.4 0*7 0*4 0*9.5

−

(U) 0*0.4 0*0.4 0*7 0*4 0*9.5

Goal of improvement (L) 0.21 0.19 3.76 1.50 3.38

Goal of improvement (U) 0.22 0.20 3.99 1.60 7.58

Goal of actual improvement (

) 0.78 0.80 6.01 3.40 12.42

Goal of actual improvement (

) 0.79 0.81 6.25 3.50 16.62

ICEIS 2008 - International Conference on Enterprise Information Systems

248