Expert Systems with Applications 42 (2015) 6050–6064

Contents lists available at ScienceDirect

Expert Systems with Applications

journal homepage: www.elsevier.com/locate/eswa

Leanness assessment and optimization by fuzzy cognitive map

and multivariate analysis
Ali Azadeh a, Mansour Zarrin a, Mohammad Abdollahi b,⇑, Saeid Noury a, Shabnam Farahmand a
a
School of Industrial Engineering, and Center of Excellence for Intelligent-Based Experimental Mechanic, College of Engineering, University of Tehran, Tehran, Iran
b
Department of Industrial and Systems Engineering, College of Engineering, Wayne State University, Detroit, MI, USA

a r t i c l e i n f o a b s t r a c t

Article history: The strategy of organizational lean production emphasizes on increasing efficiency, quality improvement
Available online 15 April 2015 and cycle time reduction by eliminating non-value added activities (MUDA). This paper presents a com-
prehensive approach based on data envelopment analysis (DEA), fuzzy DEA (FDEA), fuzzy cognitive map
Keywords: (FCM), Decision Making Trial and Evaluation Laboratory (DEMATEL) and Analytic Hierarchy Process
Lean production (AHP) for evaluating and optimizing the leanness degree of organizations to survive in competitively
Leanness assessment growing market. In this regard, a comprehensive list of quantitative and qualitative leanness measures
Fuzzy cognitive map (FCM)
is extracted from the literature. The efficiency of organizations is assessed and optimized by DEA. A
Data envelopment analysis (DEA)
Fuzzy data envelopment analysis (FDEA)
heuristic algorithm is proposed to obtain a full ranking of leanness levels of organizations. Accordingly,
Decision Making Trial and Evaluation a sensitivity analysis is carried out to determine impact of each leanness factor on lean strategy. The
Laboratory (DEMATEL) approach has been found fruitful while applying for a number of packing and printing organizations,
Analytic Hierarchy Process (AHP) in Iran, as a case study. Apart from evaluating overall lean performance metric, the proposed approach
can evaluate the impact degree of leanness factors on each other as well as the impact of leanness factors
on lean strategy. The result show that production procedure among the leanness measures has the most
impact on leanness strategy in the organizations under study. To the best of our knowledge, this is the
first study that develops and implements an efficient decision-making procedural hierarchy to support
leanness extent evaluation.
Published by Elsevier Ltd.

Motivation and significance 1. Introduction

To survive in today’s competitive market, providing products
and services with highest quality and lowest cost are the most
important challenges. During the last years of World War II many
approaches in different fields have been developed to achieve
these aims. But one of the most common approaches in the past
two decades that have paid much attention to these concerns, is
the lean manufacturing approach (Lewis, 2000). Consequently
these approaches can be utilized to evaluate a company’s success
level in implementing lean culture in comparison with other sim-
ilar organizations. Nowadays, several techniques have been devel-
oped for this aim, but these techniques have shortcomings. In this
paper, a novel approach is proposed to cope with these shortcom-
ings. The shortcomings can be expressed in two areas: first, the
proposed techniques are usually used for specific parts of organiza-
tion; second, these methods are not quite systematic.
⇑ Corresponding author.
E-mail addresses: aazadeh@ut.ac.ir (A. Azadeh), mansour.zarrin@gmail.com
(M. Zarrin), Abdollahi@wayne.edu (M. Abdollahi), saeidnoury@yahoo.com
(S. Noury), shabnamsadr@ut.ac.ir (S. Farahmand).
Lean production is introduced by Womack (1990). In an inte-
grated system, lean production is a multi-dimensional method that
includes a widespread range of management practices like just-in-
time (JIT) (Huson & Nanda, 1995), total quality management (TQM)
(Andersson, Eriksson, & Torstensson, 2006), team working
(Delbridge, Lowe, & Oliver, 2000), cellular manufacturing (Singh,
Garg, & Sharma, 2010), supplier involvement (MacDuffie &
Helper, 1997). In fact, in all industries, there is a hidden factory that
produces defective parts. Modifying processes of this factory
reduces costs of system (Miller & Vollmann, 1985).
In this context, fourteen principles of Toyota associated with
lean manufacturing are commonly used (Morgan & Liker, 2006).
According to these principles wastes is classified into three groups:
MUDA, MURI and MORA (Rinehart, 1997). MUDA refers to those
activities of processes that do not add value (Waste). MURI refers
to any variation leading to unbalanced situations (unevenness).
MORA refers to all activities asking material, employees or equip-
ment for irrational stress or effort (overburden). The elimination
of MUDA (waste) in lean manufacturing has special effects on the
performance of different industries (Rother & Shook, 2003). There

http://dx.doi.org/10.1016/j.eswa.2015.04.007
0957-4174/Published by Elsevier Ltd.
 A. Azadeh et al. / Expert Systems with Applications 42 (2015) 6050–6064
are many procedures by which these concepts can be applied.
According to lean production attitude, when non-value added
activities are eliminated, organization would benefit lean out-
comes (Wan & Chen, 2008).
One of the concerns of any organization is inventory control.
From lean production perspective, storing excessive inventory is
wrong because: (1) it increases system costs and (2) problems
and gaps of system would remain hidden (Hofer, Eroglu, &
Rossiter Hofer, 2012). Examining complexity of products design
and dynamic markets can result in valid decision variables for
transferring from traditional manufacturing to lean manufacturing
(Liker, 1997). Characteristics of lean manufacturing are reducing
costs, increasing speed of process, and reducing delivery time
(Barber & Tietje, 2008).
According to literature, only a small number of organizations’
activities are value added and unfortunately, most of processes
are non-value added (Hines & Rich, 1997) which leads to lost in time
and capital (Holweg, 2007). Consequently, two issues arise here:
(1) How to reach lean production?
(2) How to assess leanness efficiency degree in organizations?
In this paper, we aim to answer second question and use three
methods including DEA, FDEA and FCM to examine important mea-
surements. Therefore, with the help of proposed model, organiza-
tions’ status is determined in terms of leanness. Need for a novel
methodology to determine performance of a company can be
addressed by leanness approach, since the existing techniques typ-
ically focus on a specific aspect of performance and cannot show
the full range of leanness strategy (Wan & Chen, 2008).
Fig. 1. Flowchart of the proposed approach for leanness assessment.
6051
In this study, a comprehensive list of measures is obtained for
leanness assessment from the literature. To assess and evaluate
leanness level in 40 organizations, two methods consisting of
FDEA and weighted FCM are utilized and the results of the methods
are compared with each other. For verification and validation the
results of FDEA, DEA is employed and its ranking results is com-
pared with FDEA at a-cut = 1. Note that we employed two fuzzy
methods in order to alleviate vagueness and uncertainty of data
sets. In addition, DEMATEL methodology is used to assess the
degree of influences that each leanness factor has on other factors.
Also, leanness factors are ranked according to their impact degree
on lean manufacturing policies by AHP and DEA. Final ranking of
organizations is performed by a novel heuristic approach. A sensi-
tivity analysis on DEA model is performed for determining the
impact of each leanness factor on leanness strategy. Moreover,
AHP is used to consider expert’s experiments as well as to deter-
mine and rank impact level of each leanness factor on leanness
strategy. Fig. 1 depicts the roadmap of this study.
The rest of paper is organized as follows: literature review is
provided in Section 2. Problem is defined in Section 3. Model and
techniques used in this paper are briefly discussed in Section 4.
Case study is addressed in Section 5. Section 6 represents results
and sensitivity analysis. Finally, we conclude this paper in
Section 7. Future research directions are included in this section, too.
2. Literature review
The performance measurement is one of the most important
issues in manufacturing and service systems. Most managers are
6052 A. Azadeh et al. / Expert Systems with Applications 42 (2015) 6050–6064
interested to know these measures in order to have more control
over production processes. All of the models presented in this con-
text focus on quantitative measures or qualitative indicators, and
do not pay much attention to the importance of providing proper
leanness attitude.
Some of authors like (Barla, 2003; Comm & Mathaisel, 2005;
Motwani, 2003; Saurin & Ferreira, 2009; Simpson & Power, 2005;
Taj & Berro, 2006; Womack, 1990) focused on the concept of lean
manufacturing and its benefits at manufacturing organizations.
However, little work has been done in the area of lean assessment
for any type of manufacturing system (Singh et al., 2010). Soriano-
Meier and Forrester (2002) discussed integration of TQM, just in
time (JIT) and computer controlled processes with lean production
to eliminate waste. Allway and Corbett (2002) defined leanness in
terms of efficiency and effectiveness. Bayou and De Korvin (2008)
compared degree of leanness of General Motors and Ford Motor
Company. In their paper fuzzy logic was used to assess both com-
panies performance. According to this study, leanness level of pro-
duction systems in the 3-year period for Ford was 17% leaner than
the General Motors.
Another technique used to evaluate leanness is dynamic evalu-
ation. This technique uses different patterns of indicators. In fact,
the method requires range of patterns of different states of system.
Another way to assess leanness is to determine an alternative dis-
tance between current situation and best practice of system in a
comparative evaluation in order to calculate leanness level
(Srinivasaraghavan & Allada, 2006b). Swamidass (2007) utilized
ratio of inventory to value of sales as a performance criterion to
evaluate over 14,000 firm-years of leanness. It is an easy task to
analysis and interpret results using this ratio (the lower the better),
but it emphasizes only on inventory-related performance. Wan
and Chen (2008) proposed a unit-invariant leanness measure along
with self-contained benchmark to quantify level of leanness in
manufacturing industries. Applying data envelopment analysis
(DEA) model, leanness measure extracts value-adding savings from
a manufacturing process to determine leanness frontier as a bench-
mark. Azevedo, Govindan, Carvalho, and Cruz-Machado (2012)
developed Agilean index to evaluate leanness of individual organi-
zations in automotive industry. Delphi technique was used to
acquire experts’ opinions for a set of weighted agile measurements.
They conducted a comparison using defined Agilean index.
Katayama and Bennett (1999) surveyed three important concepts
for manufacturing management namely agile manufacturing,
adaptable production and lean production for organizations’ strat-
egy and performance measurements. They suggested applying
these concepts for getting adaptability results toward competitive
financial markets.
Recent studies in the field of using a linear programming model
are based on slack (SBM). But providing such models, requires
heavy computation and spends a long time (Wan & Chen, 2008).
Dahlgaard and Dahlgaard-Park (2006) analyzed results and princi-
ples of lean manufacturing. They compared three concepts consist-
ing of lean production, quality process of six sigma and total quality
management (TQM). They also discussed how to create a culture of
accepting these concepts in company. Srinivasaraghavan and
Allada (2006a) proposed a new complementary method to evaluate
leanness by providing a numerical measure and benchmarking
leanness of other organizations. They called their method as
Mahalanobis Taguchi Gram Schmidt System (MTGS).
Vinodh and Balaji (2011) designed a decision support system
based on fuzzy logic for assessing lean and utilized Euclidean dis-
tance to determine parts of organization which need modifications
and enhancements. With the help of decision support system, they
reduced error probability and saved time spent in manual compu-
tations. In another study, Vimal and Vinodh (2013) used artificial
neural networks and fuzzy logic to cope with manual computation
difficulties. They developed a simulation model for leanness
assessment and validate the model with a case study of an
Indian transformers manufacturing plant. Vimal and Vinodh
(2012) relaxed fuzzy assumption for leanness measurement by
an interface method based on IF-THEN rules. They developed a
model in a conceptual framework and determined three levels of
enabler, criterion, and attributes to determine each criterion’s
strength and suggested improvements.
Abdulmalek and Rajgopal (2007) described principles of lean
adopted in a large steel mill. They used value stream mapping
(VSM) to detect opportunities for different leanness methodolo-
gies. They also represented a simulation method to show benefits
of lean implementation such as reduced lead-time of production
and decreased numbers of work-in-process parts. Houshmand
and Jamshidnezhad (2006) used process variables, design parame-
ters and functional requirements to propose a model for process
design of lean production system based on theory of axiomatic
design. Vinodh and Chintha (2011) used a new model for leanness
measurement merged with multi-grad fuzzy method and reported
results of applying their proposed model. Seyedhosseini, Taleghani,
Bakhsha, and Partovi (2011) investigated some state-of-the arts
and implemented Balanced Scorecard (BSC) method to develop a
model for performance measurement of organizations. Leanness
criteria were selected according to BSC and DEMATEL approach
was applied to investigate the impacts of objectives on each other.
The model was able to select appropriate criteria and objectives for
an organization. Holden (2011) addressed changes occurred in sev-
eral emergency departments (EDs) after applying leanness
principles.
Petersen and Wohlin (2010) proposed a new method to take
into account practices of lean software expansion and quality
improvement. This approach is known as Software Process
Improvement by method of lean measurement. The technique
allows performance evaluation of development procedure and
determines continuous activities needed to reach a better lean
software procedure over time. Recently application of DEA in eval-
uating Decision Making Units has increased and many studies have
been done to create combined indicators (Hatefi & Torabi, 2010).
Elnadi and Shehab (2014a) presented a conceptual model that
can be used in measuring the degree of Product-Service System
(PSS) leanness in UK manufacturing companies. The model will
assess Product-Service System leanness based on five lean
enablers containing management leanness, supplier relationship,
customer relationship, process excellence, and workforce lean-
ness), and 21 criteria (process optimization, culture of manage-
ment, supplier delivery, etc.). In another study, Elnadi and
Shehab (2014b) presented an innovative model to evaluate
Product-Service System leanness incorporated with validating
the model via three UK manufacturing companies, thus improving
the practical validity and relevance of the model. The model com-
prises three levels, namely, enablers, criteria and attributes. Wu,
Xu, and Xu (2015) proposes a multiple attribute group decision
making (MAGDM) framework to facilitate the evaluation of lean
practices at logistics distribution centers. The framework deals
with the selection process and consensus process for MAGDM
problems based on the 2-tuple linguistic computation model.
Finally, the proposed framework was applied to a lean practices
evaluation problem for a commercial tobacco company’s logistics
distribution centers in China. Ram Matawale, Datta, and Sankar
Mahapatra (2014) presented an efficient fuzzy-based leanness
assessment system using generalized interval-valued trapezoidal
fuzzy numbers set. The proposed system was applied for a partic-
ular industry, in India, as a case study. They also identified ill-
performing areas towards lean achievement.
Although there are several measures for various aspects of lean
production in the literature, there is no comprehensive measure for
 A. Azadeh et al. / Expert Systems with Applications 42 (2015) 6050–6064 6053

Table 1
Comparison between features of this study and previous studies.

Study Measures of leanness Data Methodology Case study Target

This study
Bayou and De Korvin
(2008)
Behrouzi and Wong (2011)
Vinodh and Chintha (2011)
Soriano-Meier and
Forrester (2002)
Singh et al. (2010)
Wan and Chen (2008)
Jabbour, Jabbour,
Govindan, Teixeira, and
Freitas (2013)
type
See Table 2 Fuzzy Fuzzy DEA and FCM, Packing and printing Leanness assessment and
DEMATEL and AHP organizations optimization
Just-in-time (JIT), Kaizen, and Quality Fuzzy Fuzzy logic General Motors and Comparing leanness
controls Ford production system
Waste eliminations and JIT Fuzzy Final integrated unit- Manufacturing Lean performance
less score company evaluation
QFD, TQM and JIT Fuzzy Integrated fuzzy-QFD Indian electronics Enabling leanness
(manufacturing
organization)
Elimination of waste, zero defect, JIT, pull Crisp Statistical methods Manufacturing firm Evaluating leanness degree
of raw materials
Cellular manufacturing, Kaizen, Kanban, Fuzzy Human judgment error Indian auto component Measuring leanness
Material resource planning, Poka yoke industry
Value-added time Value-added cost Crisp Data envelopment Manufacturing systems Quantifying leanness level
analysis (DEA)
Total productive maintenance, Kanban, JIT, Crisp Principal component Brazilian automotive Environmental management
Kaizen, Lot reduction, Vendor analysis; statistical companies and operational
development analysis performance

overall lean implementation in the printing and packing organiza-
tions. An appropriate measurement tool is needed to evaluate the
efficiency and effectiveness of the lean implementation throughout
the entire organization which can handle the ambiguity and uncer-
tainty exist in the real systems. According to the literature survey,
we can see there are very few studies that simultaneously use DEA
model (which is an optimization model) and compare the results
with those of other methods in order to validation and verification
them. Therefore, the major contributions of this work are summa-
rized as follows: (1) a comprehensive approach is proposed to
assess leanness efficiency degree in organizations; (2) uncertain,
ambiguous and imprecise input and output data are considered
in the proposed approach; (3) DEA model which is an optimization
linear programming model (unlike statistical methods such as
regression analysis) is used in the approach; (4) the proposed
approach can evaluate and rank not only the leanness level of orga-
nizations (Step 2) but also the impact degree of leanness factors on
each other (Step 3) as well as the impact degree of leanness factors
on lean production strategy (Step 4); (5) the proposed approach
can handle even incomplete or conflicting information because of
using FCM; and (6) a sensitivity analysis is applied to verify and
validate the obtained results from each step. The features of this
study are compared with previous studies and depicted in Table 1.
of supply chain efficiency measures. Many articles are published
in this area (Sánchez & Pérez, 2001; Vinodh & Vimal, 2012; Zarei,
Fakhrzad, & Jamali Paghaleh, 2011).
3.3. Measures exhibition
In this field, authors focus on benchmarking issues and practical
principles in order to control supply chain efficiency. The objective
is to reveal the gap between implementing and planning and to
help corporations to indicate potential issues and areas for
improvement.
To cope with these problems, three different approaches are
usually used: (1) conceptual models; (2) mathematical models;
(3) hybrid approaches.
In this paper, indicators implementation (the second field) is
addressed. Some indicators used to evaluate degree of leanness
are shown in Table 2. These indicators are considered in three dif-
ferent levels. The first level consists of 5 enablers (top factors) for
lean production consisting of management responsibility, manu-
facturing leanness management, employees leanness, leanness of
Table 2
Enablers of leanness (Vinodh & Chintha, 2011).

Enablers of leanness Factors of leanness

3. Problem description and assumptions
Management responsibility (E.1) The structure of organization (F.1)
As mentioned, performance evaluation is an old issue which Management nature (F.2)
Manufacturing leanness Adaptation of customer reaction (F.3)
originally dates back to the 80th century. But evaluating perfor- management (E.2)
mance of organization by lean indicators has been considered in Changing technical and business
recent decades. Typically, these kinds of problems can be catego- processes(F.4)
rized into three groups (Gopal & Thakkar, 2012): JIT flow(F.5)
Supplier Development(F.6)
Streamlining procedures(F.7)
3.1. Measures design Cellular manufacturing (F.8)
Employees leanness (E.3) Worker status (F.9)
Worker involvement(F.10)
In this field, researchers focus on improvement of conceptual
Leanness of technology (E.4) Manufacturing setups(F.11)
models, frameworks, supported theories, and measures classifica- Product Service (F.12)
tion. Also, design of lean performance measures for improving total Integrated product design(F.13)
lean performance has been taken into account. In-house technology(F.14)
Production procedure(F.15)
Manufacturing Planning(F.16)
3.2. Measures execution Manufacturing leanness strategy Quality status (F.17)
(E.5)
In this field, survey and case study examples, and empirical Productivity status (F.18)
Cost management(F.19)
testing of frameworks have been considered by different authors.
Management of time (F.20)
The main purpose of this area is to comprehend execution issues
6054 A. Azadeh et al. / Expert Systems with Applications 42 (2015) 6050–6064
technology, and manufacturing leanness strategy (Vinodh &
Chintha, 2011). 20 lean factors are considered to assess leanness
level in organizations based on experts’ opinions.
4. Methodology description
Exchanging qualitative indicators with quantitative indicators
results in new indexes for level of being lean. To handle this issue,
three approaches including FDEA, and FCM are employed. All lean-
ness factors which are defined for one specific enabler of leanness
are merged using geometric mean of scores and are assigned to
each organization. This integration has been applied since numbers
of factors are greater than half of numbers of organizations who
wish to have their rankings. For example, factors F.1 and F.2 are
merged together; consequently numbers of factors are decreased
to 5 top factors. In this paper, to rank organizations according to
leanness degree, three methodologies are used to consider all mea-
sures and levels of priorities. Also, impact of each leanness factor
on other factors is evaluated. For this purpose, DEMATEL method-
ology is used because it utilizes experts’ opinion and determines
impact of each factor on other factors according to experts’ exper-
iments. Factors which are taken into account for leanness assess-
ment are defined based on industry classification. The results of
ranking obtained from DEA, FDEA, and FCM methods are used to
provide a full ranking of organizations. A sensitivity analysis on
DEA model is performed for determining impact of each leanness
factor on leanness strategy. In addition, Analytic Hierarchy
Process (AHP) is used to consider expert’s experiments as well as
to determine and rank impact level of each leanness factor on lean
strategy. The methodologies used in this paper are briefly
described below.
The proposed approach uses DEA model in steps 2 and 4 for
evaluation of organizations’ leanness level and assessment of the
leanness factors, respectively. Let us consider methodology where
the evaluation of lean production efficiency is based on optimiza-
tion algorithms. This includes so called multi-attribute, eco-
efficiency models, and multi-criteria decision making models
(Jablonsky, 2007). DEA models also belong to this class (Charnes,
Cooper, & Rhodes, 1978). DEA is a linear programming model to
maximize the efficiency of multiple decision-making units
(DMUs) when the production process presents a structure of mul-
tiple inputs and outputs. In DEA model, one need not assume a pri-
ori the existence of a particular production function for aggregating
and weighting inputs and outputs variables. Instead, the respective
aggregation weights result from solving an optimization problem
(Azadeh, Ghaderi, Mirjalili, & Moghaddam, 2011; Rickards, 2003).
Hence, they are solely dependent on the empirical observations
involved. This fact gives the DEA model a decisive advantage over
ordinary optimization procedures (Lee, Hsu, Chou, & Guo, 2011).
Even if one does not value the inputs and outputs variables at mar-
ket prices, or indeed in money, one still can execute the necessary
aggregations (Rickards, 2003). The objective of this paper is to opti-
mize lean performance of packing and printing organizations. By
employing the DEA model, companies can monitor efficiency
scores, which provide an indication of the levels of lean production
performance. By doing this, organizations can set a strategic goal to
achieve improved its sustainability performance in both the short-
term and long-term. The proposed approach helps managers to
assess and compare their ranking in terms of leanness degree with
other similar organizations.
4.1. Fuzzy cognitive map (FCM) procedure
FCM is a fuzzy framework which is highly similar to neural net-
works (Ghaderi, Azadeh, Pourvalikhan Nokhandan, & Fathi, 2012).
This technique can be used for managing unsupervised data. FCM
is based on expert’s opinion. The simplicity of employing this tech-
nique is the main advantage of it. FCM is applied to show the world
as a set of concepts and causality between these concepts
(Vasantha Kandasamy & Smarandache, 2003). It enables experts
to represent evaluative and factual conceptions in a communicat-
ing structure and also answer to questionnaires or rapidly draw
graphs of FCM.
FCM is a fuzzy signed graph. In FCMs, there are several loops of
causal reactions. In expert systems field, there is no need to use
methods of graph search if a feedback is used. Experts use feed-
backs to draw causality relation in problems. Static behavior can
interpret a forward-evolved implication, but FCM can be consid-
ered as a dynamic approach. Synchronous FCM acts as temporal
associative memories (TAM) (Vasantha Kandasamy &
Smarandache, 2003). As previously mentioned, FCMs are directed
graphs. These graphs can be developed by several nodes and edges.
The nodes denote some concepts such as occasions, policies, etc.,
and the edges or arrows show causalities (Vasantha Kandasamy
& Smarandache, 2003). Therefore, the causal relations among con-
cepts are represented by the graph. In FCM, nodes can be consid-
ered as fuzzy sets. The arrow eij from causal concept C i to
concept C j determines causality between C i and C j . To determine
the sign of an edge, three rules are considered:
(1) The sign of edge is positive, if a decrease (or increase) in con-
cept C i leads to decrease (or increase) in concept C j . In this
status, causal relation between two concepts is considered
positive (direct relation).
(2) The sign of edge is negative, if a decrease (or increase) in
concept C i leads to increase (or decrease) in concept C j . In
this status, causal relation between two concepts is consid-
ered negative (inverse relation).
(3) If there is no relation between two concepts, there is no
causality.
In FCM, if the causalities or weights of edge take only values 0 or
1, the FCM is called simple FCM. The sign of edges in a simple
FCM are as follows:
Positive (direct) causal relation is showed by eij ¼ þ1, and neg-
ative (inverse) casual relation is determined by eij ¼ 1, and no
causality is denoted by eij ¼ 0.
The development method of the FCM not only determines the
number and type of concepts making up the FCM, assigns linguistic
weights from experts and combines them, but it continues a step
further. The linguistic weights of the FCM are transformed into
numerical weights using the methodology proposed in (Stylios,
Groumpos, & Georgopoulos, 1999). Then, an algorithm is proposed
to adjust the weights of the FCM in order to ensure that the FCM
will always converge to a fixed desired region. Actually, the FCM
could converge to a fixed point, limit cycle, or chaotic attractor,
but when FCMs are used for the decision making process, it is
desirable to converge to a region corresponding to the selection
of one decision (Dickerson & Kosko, 1993). This algorithm is used
successfully in this paper to strengthen some weight interconnec-
tions and weaken some others and it is presented in the following
paragraphs.
A quick initial approximation is defined for an expert’s opinion
or for printed causal information. Causalty matrix or adjacency
matrix in FCM approach is shown by E ¼ ½eij nn , where ‘‘n’’ denotes
numbers of concepts (Vasantha Kandasamy & Smarandache, 2003).
Based on the adjacency matrix, an expert can list cause and effect
relations between nodes. FCM determines on–off situation of node
 
at a moment, i.e. instantaneous state of node A ¼ ½aij 1n . This
process is performed according to the following algorithm:
 A. Azadeh et al. / Expert Systems with Applications 42 (2015) 6050–6064 6055

for i ¼ 1; . . . ; n The linear counterpart of nonlinear model presented in (Eq. (1))

If ai ¼ off () ai ¼ 0 is shown below.
If ai ¼ on () ai ¼ 1 X
s

end Max uk ykp

k¼1
Including feedback to FCM results in emerging cycles. In cyclic FCM, X
m

there is at least one directed cycle. If there is no directed cycle in s:t: v j xjp ¼ 1
j¼1 ð2Þ
FCM, it is called acyclic. A dynamic system can be considered as a
FCM with feedback. Stability situation of dynamic systems is known
X
s X
m
uk yki  v j xji 6 0 i ¼ 1; . . . ; n
as hidden pattern (Vasantha Kandasamy & Smarandache, 2003). In a k¼1 j¼1
dynamic system, when the stability state is an exclusive state vec-
tor, this status is called fixed point. Based on (Vasantha Kandasamy
v k ; uj P 0
& Smarandache, 2003), algorithm to implement FCM is represented A DMU is taken into account to be effective if it gains a score
in Fig. 2. of 1 and if the score is less than 1, the DMU is ineffective. The
State vector AðtÞ is frequently moved through adjacency matrix. calculations provide a maximal performance measure using
After each move, the vector changes to be non-linearly trans- piecewise linear optimization on each DMU with respect to the
formed. A model based on assessing final situation after iterations closest observation on the frontier (Charnes et al., 1978). Based
is developed. If there are a number of frequent patterns; stability is on DEA method in addition to determining efficiency and rank-
reached for system. These repeating patterns can be restraining ing of the organizations, we could identify the optimum values
cycles or stable points or disordered attractors (Ghaderi et al., of the defined indicators, the excess values of input indicators,
2012; Vasantha Kandasamy & Smarandache, 2003). lack of output indicators and the references for each
organization.
4.2. Data envelopment analysis (DEA)
4.3. Fuzzy data envelopment analysis (FDEA)
Data envelopment analysis (DEA) was first proposed by Charnes
et al. (1978). The purpose of DEA methodology is to assess the rel- Traditional DEA method can be used only for evaluating and
ative efficiency of units that are comparable and are referred to as assessing the amount of efficiency of DMU with crisp input/output
Decision Making Units (DMU) (Charnes, 1994). These DMUs are data sets. But, in the real world and our case study, values of inputs
described according to some inputs and outputs. Score of relation and outputs are often vague and fuzzy. In this section, we use fuzzy
efficiency assigned to each DMU is defined as the ratio of weighted logic to extend traditional DEA model. DEA model in the context of
sum of outputs to weighted sum of inputs (Azadeh & Alem, 2010). fuzzy data sets is proposed as (Azadeh & Alem, 2010):
Supposing that in DEA model, the numbers of DMUs are equal
X
s
to ‘‘n’’, and there are ‘‘m’’ inputs and ‘‘s’’ outputs, the score of effi- Max uk e
y kp
ciency of a DMU like ‘‘p’’ is calculated by solving the model (Eq. (1)) k¼1
that is suggested by Charnes et al. (1978): X
s
Ps s:t: v j ex jp ¼ 1
uk ykp ð3Þ
Max Pk¼1 m
k¼1
v
j¼1 j xjp Xs X
m
Ps
ð1Þ
uk e
y ki  v j ex jp 6 0
uk yki
s:t: Pk¼1
m 6 1 k ¼ 1; . . . ; s; j ¼ 1; . . . ; m; i ¼ 1; . . . ; n k¼1 j¼1
v
j¼1 j xji vk; uj P 0
v k; uk P 0
where the symbol ‘’ shows the fuzziness of parameters.
where There are various kinds of fuzzy membership functions defined
yki : is the amount of kth output produced by DMU i in fuzzy concept, but triangular functions are mostly used. So,
xji : is the amount of jth input used by DMU i , inputs and outputs related to each DMU are considered as triangu-
v k : is the weight assigned to kth output, and lar fuzzy numbers. Consequently, model proposed for crisp data
uk : is the weight assigned to jth input. sets (Eq. (2)) can be rewritten as follows:
Let e
x ij ¼ ðxm l u e m l u
ij ; xij ; xij Þ and y ij ¼ ðyij ; yij ; yij Þ, so

X
s  
Max uk ym l u
ij ; yij ; yij
k¼1
X
s  
s:t: v j xmij ; xlij ; xuij ¼ 1
k¼1 ð4Þ
Xs   X
m  
uk ym l u
ij ; yij ; yij  v j xmij ; xlij ; xuij 6 0
k¼1 j¼1

Fig. 2. Algorithm of applying weighted FCM.
vk; uj P 0
To transform fuzzy model into non fuzzy or crisp linear pro-
gramming model, different a-cuts can be used (Jahanshahloo,
Soleimani-Damaneh, & Nasrabadi, 2004). Then, model is converted
to an interval programming. There are different procedures for
comparing the intervals. In this paper, methodology proposed by
Chang and Lee (2012) is utilized. First, a-cut method is used to
transform fuzzy DEA into interval programming as follows:
6056 A. Azadeh et al. / Expert Systems with Applications 42 (2015) 6050–6064

  2 3
X
s a11    a1j    a1n
Max uk aym l m u
kp þ ð1  aÞykp ; aykp þ ð1  aÞyij 6.
6 .. . . .. . . .. 7
k¼1
6 . . . . 7 7
  6 7
X
s
A¼6    ain 7
s:t: v i axmij þ ð1  aÞxlij ; axmij þ ð1  aÞxuij ¼1 6 ai1  aij 7 ð8Þ
6. .. .. . . .. 7
k¼1 6. 7
4. . . . . 5
X
s   ð5Þ an1  anj    ann
uk aym l m u
kj þ ð1  aÞxij ; aykj þ ð1  aÞykj
k¼1 P6
where aij ¼ 6 k¼1 xkij .
1

X
s  
Step 2: Calculate the normalized initial influence matrix D
 v j axmij þ ð1  aÞxlij ; axmij þ ð1  aÞxuij 60
k¼1
Initial direct influence matrix D ðD ¼ ð½xijnn Þ could be obtained
from normalizing average of matrix A. Matrix D could be calculated
v k ; uj P 0 through Eqs. (9) and (10), in which all elements of principal diag-
Based on our methodology, the model transforms into two onal of matrix A are equal to zero.
equivalent models that the first one determines upper bound of D¼sA ð9Þ
effectiveness and the second one provides lower bound of effec-
tiveness. If inputs are equal to their minimum values and simul- 1
taneously and outputs are equal to their maximum values, s¼ ( ) ð10Þ
X
efficiency will reach its maximum possible value. In contrast, maxi aij
minimum efficiency occurs when inputs are equal to their maxi- j

mum values and at the same time, outputs are equal to their min-
Step 3: Obtain the full direct and indirect influence matrix T
imum values.
A continuous decrease in indirect effects of problems along with
X
s   the powers of D, e.g., D2 ; D3 ; . . . ; Dk and limk!1 Dk ¼ ½0nn , where
Max uk aym u
kp þ ð1  aÞyij
P P
D ¼ ½xij nn , 0 6 xij 6 1 and i xij P 0 or i xij < 1 results in the fact
k¼1
that only sum of one row or one column of matrix D equals 1. The
X
m  
s:t: v i axmij þ ð1  aÞxuij ¼1 total-influence matrix is computed as follows:
i¼1 ð6Þ
T ¼ D þ D2 þ . . . þ Dk
X
s   X
m  
uk aym u
kj þ ð1  aÞykj  v i axmij þ ð1  aÞxlij 60 ¼ DðI þ D þ D2 þ . . . þ Dk1 ÞðI  DÞðI  DÞ1
k¼1 i¼1
¼ DðI  Dk1 ÞðI  DÞ1 ; thenT ¼ DðI  DÞ1 when limk!1 Dk
v k ; uj P 0 ¼ ½0nn ð11Þ
In this paper, range of triangular fuzzy membership function
where ¼ ½t ij nn , i; j ¼ 1; 2; . . . ; n: Furthermore, the method demon-
is based on standard deviation of distribution. Values considered
strates ri i.e. degree of impact of factor i on other factors, and cj
for a-cut are 0.1, 0.3, 0.5, 0.7, 0.9 and 1. The above model is sim-
i.e. degree of influence that other factors have on factor j.
ilar to a fuzzy linear programming model with 0 < a 6 1. Note
" #
that for different values of a, there is an optimal solution. Xn

Efficiency of DMUs calculated by FDEA methodology is repre- r ¼ ðr i Þn1 ¼ t ij ð12Þ

sented in interval form. There are different methods to rank ratio j¼1 n1
of fuzzy efficiency such as a-cut approach, Hamming distance, " #0
degree of optimality, fuzzy mean, comparison function, propor- Xn

tion to the ideal linguistic method and etc. In this paper, a-cut c ¼ ðcj Þn1 ¼ ðcj Þ01n ¼ t ij ð13Þ
j¼1 1n
method is applied.
X
s   Moreover, ri þ ci determines the intensity of influences given and
Max uk aym l
kp þ ð1  aÞykp received for factor i in which ðri þ ci Þ shows importance of factor
k¼1 i. If ðri  ci Þ > 0, then factor i is affecting other factors; and if
X
m   ðr i  ci Þ < 0, factor i is being influenced by other factors.
s:t: v i axmip þ ð1  aÞxlip ¼1 Step 4: Determine a threshold and obtain the final total influ-
i¼1 ð7Þ
    ence matrix Tp
Xs X
m
uk aym l
v i axmij þ ð1  aÞxuij To isolate relationship structure of factors, we define a thresh-
kj þ ð1  aÞykj  60
k¼1 i¼1 old to filter influences of matrix T which are significant. In matrix
v k ; uj P 0 T, each factor tij provides information about how factor i affects fac-
tor j. Consequently, if all data of matrix T is used to calculate Tp, the
matrix would be too complex to show essential data for decision
4.4. DEMATEL approach making. To diminish this complexity, the decision maker determi-
nes a threshold value for degree of influence: just factors whose
The DEMATEL method is used to determine interrelations degree of influence in matrix T is higher than threshold value can
among factors to make a Network Relationship Map (NRM). The be selected for NRM. Threshold value can be determined by brain-
method can be summarized as follows: storming technique.

Step 1: Calculate initial average matrix A 4.5. Analytic Hierarchy Process (AHP)

In this step, expert k is determines degree of immediate impact AHP is used for multiple criteria decision making processes. In
of factor i on factor j, which is shown by xkij . Initial average matrix A this method, quantitative and qualitative criteria are considered.
is represented as follows: The method is based on pairwise comparisons. Decision making
 A. Azadeh et al. / Expert Systems with Applications 42 (2015) 6050–6064 6057

begins by developing a hierarchical tree. Hierarchical decision tree
shows competing alternatives and factors. Then a series of paired
comparisons is carried out. These comparisons represent weights
of each factor in order to evaluate competing alternatives in deci-
sion making. In this paper, pairwise comparison matrices are com-
pleted according to experiments of six experts with equal weights.
5. Case study
In this case study, 40 packing and printing organizations are
studied which are quite similar in terms of products and types of
raw materials. The first questionnaire is designed based on lean-
ness factors related to each enabler of leanness. Then, the first
questionnaire is distributed among the experts of 40 different
organizations. According to measures represented in Table 2, the
first questionnaire is provided to collect required data set of
FDEA and FCM methods. The second questionnaire is designed to
evaluate impact of leanness factors on each other. DEMATEL
method is used to determine impact of each leanness factor on
other ones. The third questionnaire is used to evaluate the impact
of each leanness factor on leanness and for this purpose, we
employ AHP (a method based on experience’s expert) and DEA.
Six pieces of second and third questionnaires have been distributed
among experts chosen from 40 organizations. Based on first type of
questionnaires the required data for FDEA and weighted FCM are
collected and the second type of questionnaire is distributed
among six experts to apply DEMATEL and data of third question-
naire is used to evaluate and rank leanness factors.
Data collected through first questionnaire is fuzzy with the tri-
angular membership function because of uncertainty and subjec-
tively of data. So, for running DEA model data must be
transformed into crisp data set. There are several methods to trans-
form fuzzy data into crisp. In this paper, weighted average
approach is used for defuzzification. In this method, output equals
Table 3
Reliability statistics of the questionnaire.
Cronbach’s Cronbach’s alpha based on standardized
alpha items
0.732 0.735
to the mean of triangular membership functions. Consequently, the
mean of data is used for applying DEA.
For using DEA or FDEA, inputs and outputs must be determined.
Input variables are factors which should be minimized and output
variables are factors that should be maximized. All leanness factors
defined in Table 2 must be maximized. Therefore, DEA model tries
to maximize all output variables (all factors) considering the fact
that the input variables summation should be equal to 1 (Input-
oriented DEA model).
5.1. Reliability and consistency of data
Cronbach’s alpha is the most commonly used measure of inter-
nal consistency degree. Cronbach’s alpha is used to determine that
the scale of questions is reliable. In this paper the Cronbach’s
alphas is calculated for first questionnaire using SPSS v.22. Result
is shown in Table 3.
Note that for the third questionnaire which is a pairwise com-
parison matrix (PCM), Inconsistency Rate (IR) for each expert is cal-
culated, the values of IRs for all PCMs are lower than 0.1. The values
of IR for experts 1 to 6 are 0.01, 0.01, 0.03, 0.04, 0.01 and 0.08,
respectively.
5.2. DEA, FDEA and weighted FCM
To rank the organizations, DEA and FCM models based on the
data obtained from first type of questionnaire, are used and results
are reported in Table 4. Results of ranking using FDEA at different
levels of a-cuts are shown in Table 5. Note that, the five top factors
are taken into account for ranking process in each model.
In this case study, six experts have individually completed the
initial direct influence matrix. So, we have six 20  20 matrices
that show impact of each leanness factor on other ones. Average
of these matrices is calculated according to Step 1 of DEMATEL.
Matrix A is shown in Table 6. Then, matrices D, T and Tp are calcu-
lated respectively.
6. Results and analysis
No. of
items Three types of analysis have been done in this section. First,
20 Spearman Correlation Coefficient is used to determine correlation
between ranking results obtained from each method in order to

Table 4
Ranking results obtained from the FCM and DEA models.

Organization # FCM DEA Organization # FCM DEA

Score Rank Score Rank Score Rank Score Rank
1 4.7923 23 0.971686 19 21 4.6621 26 1.057692 6
2 4.9124 19 0.993865 16 22 5.8338 5 1.053571 7
3 4.5545 29 0.938462 30 23 5.5445 7 1.073171 4
4 4.4588 32 0.9 36 24 5.2039 11 0.9625 25
5 4.6638 25 0.971154 21 25 4.4334 33 0.9375 31
6 3.6202 40 0.804511 40 26 4.8684 22 1 14
7 4.9605 18 1.040462 10 27 5.1076 14 1.035714 11
8 3.9767 37 0.85 38 28 5.1279 13 1.076923 2
9 5.0226 17 0.933966 32 29 6.0752 1 1.043253 9
10 3.9049 38 0.910569 35 30 5.2414 9 0.971429 20
11 6.0493 2 1.019101 13 31 5.1063 15 1.058824 5
12 4.5955 28 0.968421 22 32 4.894 20 0.947761 28
13 3.6514 39 0.84127 39 33 5.1853 12 1.076923 3
14 5.1062 16 0.98209 18 34 4.5238 31 0.914286 34
15 5.6427 6 0.996985 15 35 5.9493 3 1.031742 12
16 4.3901 34 0.965517 23 36 5.3943 8 0.990392 17
17 4.5452 30 0.897135 37 37 4.6419 27 0.95 27
18 5.9436 4 1.078947 1 38 4.3672 35 0.963303 24
19 4.8862 21 0.942857 29 39 5.225 10 1.051903 8
20 4.2041 36 0.95122 26 40 4.6925 24 0.917969 33
6058 A. Azadeh et al. / Expert Systems with Applications 42 (2015) 6050–6064

Table 5
Result of ranking using FDEA model at different values of a-cut.

Org. # a-cut
0.1 0.3 0.5 0.7 0.9
Score Rank Score Rank Score Rank Score Rank Score Rank
1 0.8052 22 0.8248 22 0.8470 21 0.8726 21 0.9021 21
2 0.8513 16 0.8732 16 0.8981 16 0.9272 16 0.9623 16
3 0.7710 28 0.784 28 0.8 28 0.8169 28 0.8358 30
4 0.6468 37 0.6567 37 0.6681 37 0.6812 37 0.6965 37
5 0.8197 19 0.8381 19 0.8586 19 0.8819 19 0.9084 19
6 0.4461 40 0.4426 40 0.4385 40 0.4339 40 0.4285 40
7 0.8846 10 0.9054 10 0.9285 10 0.9545 10 0.9838 10
8 0.5735 38 0.5781 38 0.5833 38 0.5892 38 0.5961 38
9 0.7253 32 0.7392 32 0.7558 32 0.7749 32 0.7971 32
10 0.6992 35 0.7104 35 0.7229 35 0.7370 35 0.7530 35
11 0.8647 14 0.8880 14 0.9146 14 0.9450 14 0.9802 14
12 0.7960 24 0.8154 24 0.8375 24 0.8629 24 0.8924 23
13 0.5568 39 0.5595 39 0.5625 39 0.5657 39 0.5694 39
14 0.8207 18 0.8422 18 0.8669 18 0.8953 18 0.9285 18
15 0.8589 15 0.8814 15 0.9068 15 0.9363 15 0.9717 15
16 0.8082 20 0.8260 21 0.8461 22 0.8688 22 0.8947 22
17 0.6672 36 0.6779 36 0.69 36 0.7037 36 0.7196 36
18 0.8764 11 0.9009 11 0.925 11 0.9521 11 0.98295 11
19 0.7571 29 0.7727 29 0.7903 29 0.8103 31 0.8333 31
20 0.7840 26 0.7976 26 0.8125 26 0.8289 26 0.8525 26
21 0.8977 3 0.9166 2 0.9375 2 0.9605 3 0.9861 2
22 0.8977 4 0.9166 3 0.9375 3 0.9605 4 0.9861 3
23 0.8977 5 0.9166 4 0.9375 4 0.9605 2 0.9861 4
24 0.7863 25 0.8035 25 0.8229 25 0.8449 25 0.8723 25
25 0.7840 27 0.7976 27 0.8125 27 0.8289 27 0.8472 27
26 0.8977 2 0.9166 5 0.9375 5 0.9605 5 0.9861 5
27 0.8846 7 0.9054 7 0.9285 7 0.9545 7 0.9838 7
28 0.8846 8 0.9054 8 0.9285 8 0.9545 8 0.9838 8
29 0.8774 12 0.8992 12 0.9236 12 0.9511 12 0.9825 12
30 0.8070 21 0.8262 20 0.8480 20 0.8728 20 0.9028 20
31 0.9081 1 0.9255 1 0.9444 1 0.9651 1 0.9878 1
32 0.7538 30 0.7704 30 0.7894 30 0.8113 29 0.8367 29
33 0.8846 9 0.9054 9 0.9285 9 0.9545 9 0.9838 9
34 0.7 34 0.7121 34 0.7258 34 0.7413 34 0.7592 34
35 0.8676 13 0.8906 13 0.9166 13 0.9464 13 0.9807 13
36 0.8484 17 0.8682 17 0.8912 17 0.9201 17 0.9536 17
37 0.7513 31 0.7683 31 0.7878 31 0.8104 30 0.8368 28
38 0.8014 23 0.8195 23 0.84 23 0.8632 23 0.8899 24
39 0.8915 6 0.9113 6 0.9333 6 0.9577 6 0.9850 6
40 0.7105 33 0.7236 33 0.7384 33 0.7553 33 0.7748 33

Table 6
The initial average matrix A.

Factor # 1 2 3 4 5 6 7 8 9 10 11 12 F3 14 15 16 17 18 19 20
1 0 4 3 3 3 2 2 3 2 3 3 4 4 3 4 3 3 3 4 3
2 3 0 2 3 3 3 3 3 3 4 3 3 3 3 4 3 4 3 4 2
3 3 3 0 3 3 3 3 3 3 3 3 3 4 2 3 3 3 3 3 3
4 3 2 2 0 3 3 3 3 3 4 3 3 2 3 3 3 3 3 3 3
5 4 3 3 3 0 3 3 2 3 3 3 2 4 3 3 3 3 3 2 3
6 3 3 2 3 3 0 2 3 3 3 3 4 3 3 3 2 2 3 3 2
7 3 3 4 3 3 3 0 3 3 2 4 2 4 3 3 4 3 3 3 3
8 3 3 2 3 2 3 3 0 3 3 3 3 3 3 3 3 2 3 3 3
9 4 3 3 3 3 3 2 4 0 3 3 3 4 3 3 3 3 4 4 2
10 3 3 3 3 3 3 3 3 4 0 4 4 3 3 3 4 3 3 3 3
11 4 3 3 2 3 4 3 3 3 3 0 3 3 4 4 3 3 2 3 3
12 3 3 4 2 3 3 3 3 3 2 3 0 2 3 3 3 4 3 3 3
13 3 3 3 3 3 4 3 2 2 4 3 4 0 3 3 4 3 3 2 3
14 4 3 3 3 3 3 2 3 3 4 3 4 4 0 2 3 3 3 2 4
15 3 3 3 4 3 3 2 3 3 3 2 4 4 3 0 3 3 3 3 3
16 3 3 2 2 4 2 3 3 4 2 2 3 3 4 3 0 2 3 3 2
17 3 3 3 4 4 4 4 3 4 3 4 3 2 4 8 3 0 3 3 3
18 3 4 3 3 4 2 3 3 3 4 3 3 3 3 3 3 3 0 2 3
19 2 3 3 3 3 3 3 3 4 4 3 3 3 3 3 2 4 3 0 3
20 4 3 3 3 3 2 3 3 3 3 3 3 3 3 3 3 3 2 3 0
 A. Azadeh et al. / Expert Systems with Applications 42 (2015) 6050–6064 6059

Table 7
The Spearman Correlation Coefficients for performance scores.

FCM DEA FDEA (a = 0.1) FDEA (a = 0.3) FDEA (a = 0.5) FDEA (a = 0.7) FDEA (a = 0.9)
FCM 1
DEA 0.780 1
FDEA (a = 0.1) 0.749 0.949 1
FDEA (a = 0.3) 0.754 0.948 1.000 1
FDEA (a = 0.5) 0.759 0.946 0.999 1.000 1
FDEA (a = 0.7) 0.765 0.943 0.998 0.999 1.000 1
FDEA (a = 0.9) 0.770 0.938 0.996 0.997 0.999 1.000 1

Table 8
The Spearman Correlation Coefficients for ranking scores.

FCM DEA FDEA (a = 0.1) FDEA (a = 0.3) FDEA (a = 0.5) FDEA (a = 0.7) FDEA (a = 0.9)
FCM 1
DEA 0.760 1
FDEA (a = 0.1) 0.689 0.958 1
FDEA (a = 0.3) 0.699 0.963 0.999 1
FDEA (a = 0.5) 0.701 0.964 0.998 1.000 1
FDEA (a = 0.7) 0.703 0.965 0.997 0.999 0.999 1
FDEA (a = 0.9) 0.704 0.966 0.996 0.997 0.998 0.998 1

validate efficiency of methodologies. Second, a heuristic algorithm Table 9

Results of full ranking obtained from the heuristic algorithm.
is applied to present a full ranking of organizations. Finally, the
results of DEA model are analyzed to determine set of organiza- Org. # Score Rank Org. # Score Rank
tions that should be compared with each other and also to recog- 1 0.864526 21 21 0.914314 15
nize organizations defined as a template. 2 0.896824 17 22 0.97733 3
3 0.82197 30 23 0.967512 6
4 0.757977 36 24 0.874717 20
6.1. Correlation coefficient 5 0.860531 22 25 0.819189 31
6 0.597063 40 26 0.90781 16
Here Spearman Correlation Coefficient between both types of 7 0.923298 13 27 0.929902 12
results of models is presented (Spearman, 1910). Results of perfor- 8 0.684323 38 28 0.943747 8
9 0.831782 28 29 0.98416 2
mance scores and ranking scores are shown in Tables 7 and 8,
10 0.750138 37 30 0.889184 19
respectively. 11 0.973143 4 31 0.940621 11
12 0.849363 23 32 0.841543 24
13 0.655662 39 33 0.946897 7
6.2. Heuristic algorithm for full ranking of organizations
14 0.892813 18 34 0.786732 34
15 0.941027 10 35 0.972054 5
At this stage, a heuristic algorithm has been proposed to pro- 16 0.839251 26 36 0.919741 14
vide a full ranking of organization. First, we should select the best 17 0.769607 35 37 0.828104 29
FDEA model at a given level of a-cut. In this regard, the preferred 18 0.988293 1 38 0.835375 27
19 0.839263 25 39 0.942451 9
FDEA model (or best level of a-cut) is determined based on corre- 20 0.810846 32 40 0.801941 33
lation coefficients calculated for FDEA models at five different
levels of a-cut and DEA and FCM models. Selected model is a model
that has the maximum correlation coefficient with other two mod-
els. Thus, based on tests carried out, it is shown that FCM, DEA and 6.4. Final total influence matrix Tp
FDEA methods have the highest correlation coefficients with each
other at a-cut = 0.7. The performance scores of these three models Final total influence matrix Tp is computed after calculating
are normalized (i.e., all performance values are converted values matrix D and threshold value, and network relationship matrix is
between 0 and 1). Then, the average normalized performance designed according to Tp (see Table 11). Based on the network,
scores are obtained from DEA and FCM and FDEA (at a-cut = 0.7) causality relation between all leanness factors are determined.
for each organization are calculated. Therefore, a full ranking of Note that threshold value in this paper equals to the average of
organizations is provided based on the average values. The results all elements of matrix T (Threshold = 0.3173).
of this algorithm are presented in Table 9.
6.5. AHP and DEA results for ranking the leanness factors
6.3. Analysis of the DEA model
In this case study, the matrix of paired comparisons of the
In this part of analysis, the fact that each organization how and impact of various factors on lean manufacturing policy is com-
how much influence other organizations rank score has been pleted by six experts who have enough experience and knowledge
determined. In DEA model, an efficient frontier is calculated about the production system. The paired comparison matrices
(Azadeh, Amalnick, Ghaderi, & Asadzadeh, 2007). Each organiza- completed by each expert are integrated with each other through
tion that has not been recognized as an efficient organization is geometric mean. In this case is assumed that experts have equal
compared with the closest organization that is located in efficiency weights. Result of integration is shown in Table 12.
frontier (Azadeh, Ghaderi, Mirjalili, & Moghaddam, 2010). Results After integrating comparison matrix, the normalized matrix is
of this analysis are presented in Table 10. calculated, and weight of each index is obtained, and then they
6060 A. Azadeh et al. / Expert Systems with Applications 42 (2015) 6050–6064

Table 10
The impact of efficient organizations on closest non-efficient organizations.

Org. # (DMU) 7 11 18 21 22 23 26 27 28 29 31 33 35 39
1 0.203 0.052 0.19 0.554
2 0.428 0.07 0.328 0.174
3 0.077 0.923
4 0.227 0.364 0.364 0.045
5 0.744 0.256
6 0.103 0.138 0.724 0.034
8 0.125 0.875
9 0.215 0.137 0.391 0.172 0.086
10 0.447 0.046 0.338 0.169
12 0.041 0.703 0.257
13 0.966 0.034
14 0.165 0.391 0.444
15 0.42 0.478 0.042 0.06
16 0.708 0.292
17 0.066 0.081 0.458 0.395
19 0.63 0.37
20 0.667 0.292 0.042
24 0.067 0.167 0.167 0.6
25 1
30 0.518 0.424 0.059
32 0.1 0.2 0.7
34 0.4 0.6
36 0.101 0.057 0.289 0.553
37 0.56 0.112 0.164 0.164
38 0.014 0.468 0.518
40 0.645 0.266 0.089

Table 11
Final total influence matrix.

Factor # Factor #
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
1 0.33 0.32 0.34 0.32 0.33 0.34 0.34 0.35 0.36 0.34 0.38 0.32 0.33 0.33 0.33
2 0.34 0.32 0.33 0.33 0.33 0.35 0.33 0.33 0.34 0.33 0.37 0.32 0.34 0.33 0.33
3 0.32 0.32 0.33 0.33 0.32 0.34 0.35
4 0.33 0.32 0.33 0.32 0.34
5 0.33 0.32 0.34 0.35
6 0.32 0.33
7 0.34 0.33 0.32 0.33 0.33 0.33 0.34 0.33 0.34 0.32 0.36 0.33 0.36 0.33 0.33 0.33 0.32
8 0.33
9 0.34 0.32 0.33 0.33 0.32 0.33 0.34 0.32 0.36 0.33 0.33 0.33
10 0.34 0.33 0.33 0.34 0.32 0.32 0.35 0.34 0.35 0.34 0.33 0.36 0.33 0.33 0.32
11 0.34 0.32 0.32 0.33 0.33 0.32 0.33 0.33 0.36
12 0.33 0.32 0.32 0.32 0.33 0.32 0.33 0.32 0.35 0.33
13 0.33 0.33 0.33 0.32 0.35 0.33 0.32
14 0.35 0.33 0.32 0.33 0.34 0.35 0.33 0.34 0.35 0.35 0.33 0.32 0.33
15 0.34 0.33 0.32 0.32 0.33 0.33 0.34 0.35 0.35 0.32 0.32 0.32 0.32 0.32
16 0.00 0.32
17 0.37 0.36 0.34 0.36 0.38 0.36 0.35 0.35 0.38 0.38 0.38 0.37 0.37 0.37 0.48 0.36 0.36 0.36 0.33
18 0.32 0.32 0.33 0.32 0.34 0.32 0.33 0.34 0.35
19 0.33 0.32 0.33 0.34 0.32 0.32 0.33 0.36 0.35 0.33 0.34 0.35 0.34 0.37 0.32 0.35 0.33
20 0.33 0.32 0.33 0.34

are ranked. Also, the impact of each leanness factor on lean manu-
facturing strategy is calculated by DEA. For determining the impact
of each factor, two ranking methods are used; AHP and DEA. AHP is
based on expert’s opinion and DEA is a mathematical model.
Ranking results of two methods are presented in Table 13.
According to results of Table 13, the reader can see that production
procedure (F.15) has the most impact on leanness strategy in the
organizations under study. Also, this factor (F.15) has received
the first place in two methods.
6.6. Sensitivity analysis of DEA model
In this part, the DEA model is analyzed. For this purpose,
different groups of top leanness factors are selected according
to experts’ opinion and results of AHP. For each group of top
leanness factors, DEA model is run and then, the results of each
run DEA are compared with the results of main DEA model that
considers all top leanness factors (i.e. F.1, F.2, F.3, F.4 and F.5).
Then, the correlation coefficient among ranking scores of each
group and ranking scores of main model is computed to deter-
mine groups of top factors that have greater impact on main
ranking model. Top factors that have more impact on leanness
strategy are determined through this analysis. Results are repre-
sented in Table 14. According to Table 14, group 2 has the most
correlation coefficient with main model results among the
groups considering three top factors. Therefore, top factors 1,
2 and 3 have the most impact on leanness strategy in the
organizations.
 A. Azadeh et al. / Expert Systems with Applications 42 (2015) 6050–6064 6061

Table 12
The integrated paired comparison matrix of leanness factors.

Factor # 1 2 3 4 5 6 7 8 F.9 10 11 12 13 14 15 16 17 18 19 20
1 1.00 1.26 1.07 0.76 1.07 1.26 0.95 0.76 0.76 0.89 0.90 1.00 0.85 0.95 0.53 1.35 1.03 0.74 1.16 1.05
2 0.79 1.00 0.85 0.60 0.85 1.00 0.76 0.61 0.61 0.71 0.71 0.79 0.67 0.76 0.42 1.07 0.82 0.58 0.92 0.83
3 0.93 1.18 1.00 0.71 1.00 1.18 0.89 0.71 0.71 0.83 0.84 0.93 0.79 0.89 0.50 1.26 0.96 0.69 1.08 0.98
4 1.32 1.67 1.41 1.00 1.41 1.67 1.26 1.01 1.01 1.18 1.19 1.32 1.12 1.26 0.71 1.78 1.36 0.97 1.53 1.39
5 0.93 1.18 1.00 0.71 1.00 1.18 0.89 0.71 0.71 0.83 0.84 0.93 0.79 0.89 0.50 1.26 0.96 0.69 1.08 0.98
6 0.79 1.00 0.85 0.60 0.85 1.00 0.76 0.61 0.61 0.71 0.71 0.79 0.67 0.76 0.42 1.07 0.82 0.58 0.92 0.83
7 1.05 1.32 1.12 0.79 1.12 1.32 1.00 0.80 0.80 0.93 0.94 1.05 0.89 1.00 0.56 1.41 1.08 0.77 1.21 1.10
8 1.31 1.65 1.40 0.99 1.40 1.65 1.25 1.00 1.00 1.16 1.18 1.31 1.11 1.25 0.70 1.76 1.35 0.96 1.51 1.37
9 1.31 1.65 1.40 0.99 1.40 1.65 1.25 1.00 1.00 1.16 1.18 1.31 1.11 1.25 0.70 1.76 1.35 0.96 1.51 1.37
10 1.12 1.41 1.20 0.85 1.20 1.41 1.07 0.86 0.86 1.00 1.01 1.12 0.95 1.07 0.60 1.51 1.16 0.83 1.30 1.18
11 1.11 1.40 1.19 0.84 1.19 1.40 1.06 0.85 0.85 0.99 1.00 1.11 0.94 1.06 0.59 1.50 1.14 0.82 1.28 1.16
12 1.00 1.26 1.07 0.76 1.07 1.26 0.95 0.76 0.76 0.89 0.90 1.00 0.85 0.95 0.53 1.35 1.03 0.74 1.16 1.05
13 1.18 1.48 1.26 0.89 1.26 1.48 1.12 0.90 0.90 1.05 1.06 1.18 1.00 1.12 0.63 1.59 1.21 0.87 1.36 1.24
14 1.05 1.32 1.12 0.79 1.12 1.32 1.00 0.80 0.80 0.93 0.94 1.05 0.89 1.00 0.56 1.41 1.08 0.77 1.21 1.10
15 1.87 2.36 2.01 1.42 2.01 2.36 1.79 1.43 1.43 1.67 1.69 1.87 1.59 1.79 1.00 2.53 1.93 1.38 2.17 1.97
16 0.74 0.93 0.79 0.56 0.79 0.93 0.71 0.57 0.57 0.66 0.67 0.74 0.63 0.71 0.40 1.00 0.76 0.55 0.86 0.78
17 0.97 1.22 1.04 0.73 1.04 1.22 0.92 0.74 0.74 0.86 0.87 0.97 0.82 0.92 0.52 1.31 1.00 0.71 1.12 1.02
18 1.36 1.71 1.45 1.03 1.45 1.71 1.29 1.04 1.04 1.21 1.22 1.36 1.15 1.29 0.72 1.83 1.40 1.00 1.57 1.42
19 0.86 1.09 0.92 0.65 0.92 1.09 0.82 0.66 0.66 0.77 0.78 0.86 0.73 0.82 0.46 1.16 0.89 0.64 1.00 0.91
20 0.95 1.20 1.02 0.72 1.02 1.20 0.91 0.73 0.73 0.85 0.86 0.95 0.81 0.91 0.51 1.28 0.98 0.70 1.10 1.00

Table 13 used in the proposed approach because FCMs can successfully rep-
Ranking results of leanness impact factors on lean production strategy. resent human and knowledge experience, introducing concepts to
Factors AHP DEA represent the essential elements and the cause and effect relation-
Weight Rank Efficiency Rank
ships among the concepts to model the behavior of any system.
Based on the results of DEMATEL, the causality relation between
Organization’s structure (F.1) 0.923203 11 0.947368 12
Management nature (F.2) 0.732722 19 0.9 14
all leanness factors are determined. Therefore, we can analyze
Adaptation of customer reaction (F.3) 0.862891 15 0.913978 13 and consider these relations for future planning of the organiza-
Changing technical and business 1.220281 3 0.9 18 tions. The results of DEA model were analyzed to determine set
processes (F.4) of organizations that should be compared with each other and also
JIT flow (F.5) 0.862848 16 0.964286 8
to recognize organizations defined as a template. The impact of
Supplier development (F.6) 0.732759 18 0.8 19
Streamlining procedures (F.7) 0.968538 10 0.9 15 each leanness factor on lean manufacturing strategy was calcu-
Cellular manufacturing (F.8) 1.207226 5 1 4 lated and according to the results, production procedure (F.15)
Worker status (F.9) 1.207236 4 1.1 3 has the most impact on leanness strategy in the organizations
Worker involvement (F.10) 1.03626 7 0.957447 9 under study. Finally, we selected different groups of top leanness
Manufacturing setups (F.11) 1.025139 8 1 5
factors based on experts’ opinion and results of AHP and run DEA
Product service (F.12) 0.923195 12 0.9 16
Integrated product design (F.13) 1.087147 6 0.954545 10 model for each group. By calculating the correlation coefficient
In-house technology (F.14) 0.968562 9 0.947368 11 among ranking scores of each group and ranking scores of main
Production procedure (F.15) 1.730691 1 1.312 1 model, we determined the factors which have greater impact on
Manufacturing planning (F.16) 0.684877 20 0.8 20
main ranking model in our case study. These factors were the
Quality status (F.17) 0.895586 13 1 6
Productivity status(F.18) 1.252987 2 1.133333 2 structure of organization (F.1), management nature (F.2) and adap-
Cost management (F.19) 0.797869 17 1 7 tation of customer reaction (F.3).
Management of time (F.20) 0.879982 14 0.9 17 The main contributions of the proposed approach have been
summarized as follows: implementation and development of an
effective decision-making procedural hierarchy to support lean-
Table 14
ness extent assessment; considering uncertain, ambiguous and
Correlation coefficient between each group of factors and main ranking. imprecise input and output data in proposed approach; using
the abilities and advantages of DEA model such as optimizing
No. of group Top factors Correlation with main ranking results
lean production efficiency, handling multiple inputs and outputs,
1 {F.1, F.2} 0.584 uncovering relationships that remain hidden for other method-
2 {F.1, F.2, F.3} 0.808
3 {F.1, F.3, F.4} 0.774
ologies and capable of being used with any input–output mea-
4 {F.3, F.4, F.5} 0.647 surement; an overall lean performance index evaluation
5 {F.2, F.3, F.4} 0.716 platform has been introduced that can evaluate the leanness
6 {F.1, F.4, F.5} 0.796 level of organizations, the impact degree of leanness factors on
7 {F.1, F.2, F.4} 0.687
each other and the impact degree of each leanness factor on lean
8 {F.2, F.4, F.5} 0.668
9 {F.2, F.3, F.4, F.5} 0.802 production strategy; considering the major advantage of FCMs
10 {F.1, F.3, F.4, F.5} 0.874 i.e. handling incomplete or conflicting information because the
11 {F.1, F.2, F.3, F.4} 0.804 input and output data in real-world problems may be missing,
unreliable or be difficult to integrate with other information;
verification and validation of the proposed approach by compar-
7. Conclusion and future research directions ing the results obtained from each step with those of another
method. Also, the approach was validated by applying the model
In this study, an overall lean performance index evaluation to compute the leanness level for 40 Iranian packing and print-
approach has been introduced. Concept of fuzzy numbers has been ing and identifying areas for further improvement of the
efficiently explored to facilitate this decision-making. FCMs were organizations.
6062 A. Azadeh et al. / Expert Systems with Applications 42 (2015) 6050–6064

Limitations of the proposed approach include the comprehen-

Practicability

sive nature of the tool. First, whether the organization operates

in a services or manufacturing industry may make some differ-
U ences in applying the approach, considering that some perfor-
U

U
U
U
mance dimensions contain an industrial bias in the proposed
approach. Second, the companies may prefer to give a weight
experiences

to each performance indicator. While some performance dimen-

sions may have a lower importance weight in particular organi-
expert
Using

zations, the others might be more important in other industries.

U

U
Third, the proposed model may not cover all performance
Verification

mechanism

dimensions that have a potential to assess leanness level in

validation

manufacturing organizations, but we believe it captures the most

and

important. The advantages of this study are compared with pre-

U

vious studies in this section and results are represented in

uncertainty

Table 15.
Handling

Five future directions can be addressed here. First, developing

more leanness measures might be very useful to cover a more com-
U

U
prehensive scope of organizations. Second, intelligent methods are
information

highly recommended to improve the proposed ranking procedure;

incomplete

conflicting
Handling

such as neural network implementation to adapt weights of DEA,

FDEA, and FCM for final ranking. Third, future work would include
U
or

the collection of survey data directly from the manufacturing orga-

nizations. The review can be structured according to the type of the
qualitative
Handling

analysis requirements based on one or more considerations such as

company size, geographical location of companies, and companies
data

belonging to a given sector. Even though, the metrics used to mea-

sure the overall leanness can be extended to measure the improve-
Multiple

outputs
inputs

ment in any specific area of an organization or measure

and

departmental effectiveness. Forth, to create better matching road-

maps for various systems, future researches can apply more
Optimizing

production
efficiency

advanced computer intelligence, such as agent-based systems or

swarm-based in their lean assessment model. Fifth, as the human
lean

asset is the key to long-term superior organizational performance,

U

considering the best human performance practices required to sus-

Manufacturing

tain and achieve safety, quality, and work productivity. In this

regard, the issue of work compatibility will become an important
strategy

concept that should be mathematically developed in the future

researches.
U

U
Technology

Acknowledgment
U
U

U
U
U

The authors are grateful for the valuable comments and sugges-
tions from the respected reviewers. Their valuable comments and
Employees

suggestions have enhanced the strength and significance of our

paper. This study was supported by a grant from University of
U

U
U

U
U
U

Tehran (Grant No. 8106013/1/19). The authors are grateful for

Manufacturing

the support provided by the College of Engineering, University of

management

Tehran, Iran. This study was also supported by a grant from the
Iran National Science Foundation [grant number 93010029]. The
authors are grateful for the financial support provided by the
Leanness enablers

U
U

U
U

U
U
U

U
U

U
U
U

Iran National Science Foundation.

Management

References
system

U
U
U
U
U

Abdulmalek, F. A., & Rajgopal, J. (2007). Analyzing the benefits of lean

manufacturing and value stream mapping via simulation: a process sector
Hosseini Nasab and Khademi
Sharma and Bhagwat (2007)
Cousins, Lawson, and Squire

Ram Matawale et al. (2014)

case study. International Journal of Production Economics, 107(1), 223–236.

Elnadi and Shehab (2014b)
Elnadi and Shehab (2014a)
Abdulmalek, Rajgopal, and
Pakdil and Leonard (2014)

Dües, Tan, and Lim (2013)

Klassen, and Gavronski
The advantages of this study.

Abdulmalek, F. A., Rajgopal, J., & Needy, K. L. (2006). A classification scheme for the
Hajmohammad, Vachon,
Seth⁄ and Gupta (2005)
Pampanelli, Found, and

process industry to guide the implementation of lean. Engineering Management

Jabbour et al. (2013)
Bernardes (2013)

Journal, 18(2).
Allway, M., & Corbett, S. (2002). Shifting to lean service: Stealing a page from
Needy (2006)

Wu et al. (2015)
Zare (2012)

manufacturers’ playbooks. Journal of Organizational Excellence, 21(2), 45–54.

Andersson, R., Eriksson, H., & Torstensson, H. (2006). Similarities and differences
This study

(2013)
(2008)

between TQM, six sigma and lean. The TQM Magazine, 18(3), 282–296.
Azadeh, A., & Alem, S. (2010). A flexible deterministic, stochastic and fuzzy data
Table 15

Study

envelopment analysis approach for supply chain risk and vendor selection
problem: Simulation analysis. Expert Systems with Applications, 37(12),
7438–7448.
 A. Azadeh et al. / Expert Systems with Applications 42 (2015) 6050–6064
Azadeh, A., Amalnick, M., Ghaderi, S., & Asadzadeh, S. (2007). An integrated DEA PCA
numerical taxonomy approach for energy efficiency assessment and
consumption optimization in energy intensive manufacturing sectors. Energy
Policy, 35(7), 3792–3806.
Azadeh, A., Ghaderi, S., Mirjalili, M., & Moghaddam, M. (2011). Integration of
analytic hierarchy process and data envelopment analysis for assessment and
optimization of personnel productivity in a large industrial bank. Expert Systems
with Applications, 38(5), 5212–5225.
Azadeh, A., Ghaderi, S. F., Mirjalili, M., & Moghaddam, M. (2010). A DEA approach for
ranking and optimisation of technical and management efficiency of a large
bank based on financial indicators. International Journal of Operational Research,
9(2), 160–187.
Azevedo, S. G., Govindan, K., Carvalho, H., & Cruz-Machado, V. (2012). An integrated
model to assess the leanness and agility of the automotive industry: Resources.
Conservation and Recycling, 66, 85–94.
Barber, C. S., & Tietje, B. C. (2008). A research agenda for value stream mapping the
sales process. Journal of Personal Selling and Sales Management, 28(2), 155–165.
Barla, S. B. (2003). A case study of supplier selection for lean supply by using a
mathematical model. Logistics Information Management, 16(6), 451–459.
Bayou, M. E., & De Korvin, A. (2008). Measuring the leanness of manufacturing
systems—A case study of ford motor company and general motors. Journal of
Engineering and Technology Management, 25(4), 287–304.
Behrouzi, F., & Wong, K. Y. (2011). Lean performance evaluation of manufacturing
systems: A dynamic and innovative approach. Procedia Computer Science, 3,
388–395.
Chang, P.-T., & Lee, J.-H. (2012). A fuzzy DEA and knapsack formulation integrated
model for project selection. Computers & Operations Research, 39(1), 112–125.
Charnes, A. (1994). Data envelopment analysis: Theory, methodology and applications.
Springer.
Charnes, A., Cooper, W. W., & Rhodes, E. (1978). Measuring the efficiency of decision
making units. European Journal of Operational Research, 2(6), 429–444.
Comm, C. L., & Mathaisel, D. F. X. (2005). A case study in applying lean sustainability
concepts to universities. International Journal of Sustainability in Higher
Education, 6(2), 134–146.
Cousins, P. D., Lawson, B., & Squire, B. (2008). Performance measurement in
strategic buyer-supplier relationships: The mediating role of socialization
mechanisms. International Journal of Operations & Production Management,
28(3), 238–258.
Dahlgaard, J. J., & Dahlgaard-Park, S. M. (2006). Lean production, six sigma quality.
TQM and company culture: The TQM magazine, 18(3), 263–281.
Delbridge, R., Lowe, J., & Oliver, N. (2000). Shopfloor responsibilities under lean
teamworking. Human Relations, 53(11), 1459–1479.
Dickerson, J. A., & Kosko, B. (1993). Virtual worlds as fuzzy cognitive maps. In Virtual
reality annual international symposium, 1993 (pp. 471–477). Seattle, WA: IEEE.
Dües, C. M., Tan, K. H., & Lim, M. (2013). Green as the new lean: How to use lean
practices as a catalyst to greening your supply chain. Journal of Cleaner
Production, 40, 93–100.
Elnadi, M., & Shehab, E. (2014a). A conceptual model for evaluating product-service
systems leanness in UK manufacturing companies. Procedia CIRP, 22, 281–286.
Elnadi, M., & Shehab, E. (2014b). A multiple-case assessment of product-service
system leanness in UK manufacturing companies. Proceedings of the Institution
of Mechanical Engineers, Part B: Journal of Engineering Manufacture.
0954405414555561.
Ghaderi, S., Azadeh, A., Pourvalikhan Nokhandan, B., & Fathi, E. (2012). Behavioral
simulation and optimization of generation companies in electricity markets by
fuzzy cognitive map. Expert Systems with Applications, 39(5), 4635–4646.
Gopal, P. R. C., & Thakkar, J. (2012). A review on supply chain performance measures
and metrics: 2000–2011. International Journal of Productivity and Performance
Management, 61(5), 518–547.
Hajmohammad, S., Vachon, S., Klassen, R. D., & Gavronski, I. (2013). Lean
management and supply management: their role in green practices and
performance. Journal of Cleaner Production, 39, 312–320.
Hatefi, S., & Torabi, S. (2010). A common weight MCDA–DEA approach to construct
composite indicators. Ecological Economics, 70(1), 114–120.
Hines, P., & Rich, N. (1997). The seven value stream mapping tools. International
Journal of Operations & Production Management, 17(1), 46–64.
Hofer, C., Eroglu, C., & Rossiter Hofer, A. (2012). The effect of lean production on
financial performance: The mediating role of inventory leanness. International
Journal of Production Economics.
Holden, R. J. (2011). Lean thinking in emergency departments: A critical review.
Annals of Emergency Medicine, 57(3), 265–278.
Holweg, M. (2007). The genealogy of lean production. Journal of Operations
Management, 25(2), 420–437.
Hosseini Nasab, H., & Khademi Zare, H. (2012). Finding a probabilistic approach to
analyze lean manufacturing. Journal of Cleaner Production, 29, 73–81.
Houshmand, M., & Jamshidnezhad, B. (2006). An extended model of design process
of lean production systems by means of process variables. Robotics and
Computer-Integrated Manufacturing, 22(1), 1–16.
Huson, M., & Nanda, D. (1995). The impact of just-in-time manufacturing on firm
performance in the US. Journal of Operations Management, 12(3), 297–310.
Jabbour, C. J. C., Jabbour, A. B. L. d. S., Govindan, K., Teixeira, A. A., & Freitas, W. R. d.
S. (2013). Environmental management and operational performance in
automotive companies in Brazil: The role of human resource management
and lean manufacturing. Journal of Cleaner Production, 47, 129–140.
Jablonsky, J. (2007). Measuring the efficiency of production units by AHP models.
Mathematical and Computer Modelling, 46(7), 1091–1098.
6063
Jahanshahloo, G. R., Soleimani-Damaneh, M., & Nasrabadi, E. (2004). Measure of
efficiency in DEA with fuzzy input–output levels: A methodology for assessing,
ranking and imposing of weights restrictions. Applied Mathematics and
Computation, 156(1), 175–187.
Katayama, H., & Bennett, D. (1999). Agility, adaptability and leanness: A comparison
of concepts and a study of practice. International Journal of Production Economics,
60, 43–51.
Lee, K.-M., Hsu, M.-R., Chou, J.-H., & Guo, C.-Y. (2011). Improved differential
evolution approach for optimization of surface grinding process. Expert Systems
with Applications, 38(5), 5680–5686.
Lewis, M. A. (2000). Lean production and sustainable competitive advantage.
International Journal of Operations & Production Management, 20(8), 959–978.
Liker, J. K. (1997). Becoming lean: Inside stories of US manufacturers. Productivity
Press.
MacDuffie, J. P., & Helper, S. (1997). Creating lean suppliers: Diffusing lean
production through the supply chain. California Management Review, 39(4).
Miller, J. G., & Vollmann, T. E. (1985). The hidden factory: Harvard business review,
63(5), 142–150.
Morgan, J. M., & Liker, J. K. (2006). The Toyota product development system:
Integrating people, process, and technology. Charlotte, NC: B&T.
Motwani, J. (2003). A business process change framework for examining lean
manufacturing: A case study. Industrial Management & Data Systems, 103(5),
339–346.
Pakdil, F., & Leonard, K. M. (2014). Criteria for a lean organisation: development of a
lean assessment tool. International Journal of Production Research, no. ahead-of-
print, p. 1–21.
Pampanelli, A. B., Found, P., & Bernardes, A. M. (2013). A Lean & Green model for a
production cell. Journal of Cleaner Production.
Petersen, K., & Wohlin, C. (2010). Software process improvement through the lean
measurement (SPI-LEAM) method. Journal of Systems and Software, 83(7),
1275–1287.
Ram Matawale, C., Datta, S., & Sankar Mahapatra, S. (2014). Leanness estimation
procedural hierarchy using interval-valued fuzzy sets (IVFS): Benchmarking. An
International Journal, 21(2), 150–183.
Rickards, R. C. (2003). Setting benchmarks and evaluating balanced scorecards with
data envelopment analysis. Benchmarking: An International Journal, 10(3),
226–245.
Rinehart, J. W. (1997). Just another car factory?: Lean production and its discontents.
Cornell University Press.
Rother, M., & Shook, J. (2003). Learning to see: value stream mapping to add value
and eliminate muda. Lean Enterprise Institute.
Sánchez, A. M., & Pérez, M. P. (2001). Lean indicators and manufacturing strategies.
International Journal of Operations & Production Management, 21(11),
1433–1452.
Saurin, T. A., & Ferreira, C. F. (2009). The impacts of lean production on working
conditions: A case study of a harvester assembly line in Brazil. International
Journal of Industrial Ergonomics, 39(2), 403–412.
Seth⁄, D., & Gupta, V. (2005). Application of value stream mapping for lean
operations and cycle time reduction: an Indian case study. Production Planning
& Control, 16(1), 44–59.
Seyedhosseini, S. M., Taleghani, A. E., Bakhsha, A., & Partovi, S. (2011). Extracting
leanness criteria by employing the concept of balanced scorecard. Expert
Systems with Applications, 38(8), 10454–10461.
Sharma, M. K., & Bhagwat, R. (2007). An integrated BSC-AHP approach for supply
chain management evaluation. Measuring Business Excellence, 11(3), 57–68.
Simpson, D. F., & Power, D. J. (2005). Use the supply relationship to develop lean and
green suppliers. Supply Chain Management: An International Journal, 10(1),
60–68.
Singh, B., Garg, S., & Sharma, S. (2010). Development of index for measuring
leanness: Study of an Indian auto component industry. Measuring Business
Excellence, 14(2), 46–53.
Soriano-Meier, H., & Forrester, P. L. (2002). A model for evaluating the degree of
leanness of manufacturing firms. Integrated Manufacturing Systems, 13(2),
104–109.
Spearman, C. (1910). Correlation calculated from faulty data. British Journal of
Psychology, 3(3), 271–295. 1904–1920.
Srinivasaraghavan, J., & Allada, V. (2006a). Application of mahalanobis distance as a
lean assessment metric. The International Journal of Advanced Manufacturing
Technology, 29(11–12), 1159–1168.
Srinivasaraghavan, J., & Allada, V. (2006b). Application of mahalanobis distance as a
lean assessment metric. The International Journal of Advanced Manufacturing
Technology, 29(11), 1159–1168.
Stylios, C. D., Groumpos, P. P., & Georgopoulos, V. C. (1999). Fuzzy cognitive map
approach to process control systems. Journal of Advanced Computational
Intelligence, 3(5), 409–417.
Swamidass, P. M. (2007). The effect of TPS on US manufacturing during 1981–1998:
Inventory increased or decreased as a function of plant performance.
International Journal of Production Research, 45(16), 3763–3778.
Taj, S., & Berro, L. (2006). Application of constrained management and lean
manufacturing in developing best practices for productivity improvement in an
auto-assembly plant. International Journal of Productivity and Performance
Management, 55(3/4), 332–345.
Vasantha Kandasamy, W., & Smarandache, F. (2003). Fuzzy cognitive maps and
neutrosophic cognitive maps.
Vimal, K., & Vinodh, S. (2012). Leanness evaluation using IF–THEN rules. The
International Journal of Advanced Manufacturing Technology, 63(1–4), 407–413.
6064 A. Azadeh et al. / Expert Systems with Applications 42 (2015) 6050–6064
Vimal, K., & Vinodh, S. (2013). Application of artificial neural network for fuzzy logic
based leanness assessment. Journal of Manufacturing Technology Management,
24(2), 274–292.
Vinodh, S., & Balaji, S. (2011). Fuzzy logic based leanness assessment and its
decision support system. International Journal of Production Research, 49(13),
4027–4041.
Vinodh, S., & Chintha, S. K. (2011). Leanness assessment using multi-grade fuzzy
approach. International Journal of Production Research, 49(2), 431–445.
Vinodh, S., & Vimal, K. E. K. (2012). Thirty criteria based leanness assessment using
fuzzy logic approach. The International Journal of Advanced Manufacturing
Technology, 1–11.
Wan, H.-D., & Chen, F. F. (2008). A leanness measure of manufacturing systems for
quantifying impacts of lean initiatives. International Journal of Production
Research, 46(23), 6567–6584.
Womack, J. P. (1990). Machine that changed the world. Scribner.
Wu, Z., Xu, J., & Xu, Z. (2015). A multiple attribute group decision making
framework for the evaluation of lean practices at logistics distribution centers.
Annals of Operations Research, 1–23.
Zarei, M., Fakhrzad, M., & Jamali Paghaleh, M. (2011). Food supply chain
leanness using a developed QFD model. Journal of Food Engineering, 102(1),
25–33.