statistical and other methods, it is a relatively new

EVALUATING THE nonparametric method [Gupta 2014]. DEA is an optimisation

method that is used to assess the technical efficiency,
EFFICIENCY OF LEAN performance or productivity of production units based on the
levels of inputs and outputs [Cook 2005]. While there are
MANAGEMENT PROJECTS multiple analytical tools available for calculating technical
efficiency, DEA is one of the simplest and most efficient ones
USING DATA ENVELOPMENT [Sherman 2006]. DEA makes it possible to individually assess
the efficiency of each production unit relative to the entire set
of units. The objective of this method is to classify production
ANALYSIS units as efficient or inefficient and determine how an inefficient
EVA STICHHAUEROVA, NATALIE PELLONEOVA unit can reduce its inputs or increase its outputs in order to be
considered efficient. DEA is convenient for determining the
Technical University of Liberec, Faculty of Economics technical efficiency of units that are mutually comparable
Department of Business Administration and Management [Ramanathan 2003]. Such units use the same inputs and
Liberec, Czech Republic produce the same outputs, but there are certain differences in
their performance. The units assessed are most often
DOI: 10.17973/MMSJ.2018_03_2017112
companies, organisational units, banks, hospitals, public
e-mail: [email protected] administration organisations, territorial units etc. [Kumar
This paper describes approach to evaluate and compare 2007].
efficiency of lean management projects, using a set of
The basic objective of DEA is to compare the productivity of
techniques known as Data Envelopment Analysis. DEA is a non-
organisational units, here referred to as Decision Making Units
parametric linear programming method used to the efficiency
(DMUs). The operations of each DMU require certain inputs
evaluation of decision making units. Lean methods have been
and result in certain outputs. Inputs are quantities that are
widely used as a tool for improving operational performance
consumed in a particular operation, and outputs are the
and also have been successfully implemented in many
resulting products. In general, lower input values and higher
manufacturing and non-manufacturing organizations. This
output values are preferred. Unlike common efficiency-rate
paper aims to develop a mathematical model to evaluate the
calculations, DEA uses mathematical programming that makes
efficiency of lean management projects. Paper provides the
it possible to include a large number of inputs and outputs in
identification of important inputs and outputs for projects that
the model [Kumar 2007].
are then analysed using DEA. Working with planned indicators
values, this model helps to identify one or more projects that The DEA model is used to calculate the relative technical
result in the maximum benefit to the organization. Using the efficiency score, which expresses a DMU’s efficiency within the
real indicators values after implementation of a project, it can group of DMUs under study. Relative technical efficiency can be
help to determine how the project was successful as compared defined as the ratio between total weighted production and
with similar implemented projects. total weighted consumption of inputs. Based on efficiency rate,
KEYWORDS DMUs within the group are then classified as efficient or
data envelopment analysis, lean management project, project inefficient. Given the mechanism for choosing the weights of
efficiency, project success, technical efficiency inputs and outputs, within the set of DMUs under study there is
always at least one efficient DMU [Bogetoft 2011]. This kind of
1 INTRODUCTION efficiency rate measures the DMU’s distance from the
efficiency frontier and expresses the proportional reduction of
This article aims to present the DEA method as a convenient all inputs (for input-oriented models) or the proportional
tool for evaluating and subsequently selecting projects or for increase of all outputs (for output-oriented models) that is
arranging lean management projects by priority. In this article, necessary in order to move the DMU onto the efficiency
the authors have focused on assessing the relative efficiency of frontier [Ray 2004].
20 hypothetical lean management projects, namely projects to
apply the Single Minute Exchange of Dies (SMED) method using The best-known models are CCR and BCC. Designed by Charnes,
a selected Data Envelopment Analysis (DEA) model. The Cooper and Rhodes in 1978, the CCR model is the historically
methodology presented in this study aims to help managers first DEA model. The CCR model can be either input or output
identify efficient SMED projects and rank projects according to oriented, and it has been designed under constant returns to
specified criteria, thus facilitating the subsequent selection of scale. The assumption of constant returns to scale is convenient
the project to be implemented. When applying the DEA in cases where all businesses operate at the optimal scale. As a
method, it is assumed that projects identified as efficient will result of imperfect competition, financial constraints etc. a
lead to the highest performance and, in turn, maximise utility business may not operate at the optimal scale. In 1984, Banker,
for the organisation [Kumar 2007]. Charnes and Cooper proposed an extension to the CCR model –
the BCC model. Similarly to the CCR model, the BCC model
derives from the role of mathematical programming. While the
2 LITERATURE REVIEW BCC model can also be either input or output oriented, unlike
2.1 Data Envelopment Analysis the CCR model it assumes variable returns to scale [Yao 2010].
Many approaches and techniques have been proposed in With variable returns to scale, there are three distinct areas: an
connection with addressing project selection. Data area of increasing returns to scale, an area of decreasing
Envelopment Analysis (DEA) is a method used to solve multi- returns to scale, and an area of constant returns to scale.
criteria problems and, in recent years, it has been frequently Assuming variable returns to scale, it no longer applies that – in
used to assess the efficiency of production units. DEA is an order to maintain efficiency – an α-multiple of inputs must be
important tool of economic management. Compared to matched with the same multiple of outputs. As a result of the

MM SCIENCE JOURNAL I 2018 I MARCH

2306
assumption of variable returns to scale, a DMU will be classified ranking of Lucent Technologies’ telecommunications R&D
as efficient even if the relative increase in outputs is lower or projects [Linton 2007]. A more comprehensive approach was
greater than the corresponding increase in inputs. In that case, adopted by [Eilat 2008] in assessing R&D projects at different
the technical efficiency rate of the DMUs being assessed will be stages of their life cycle. A model was used that combined the
greater (or not lower) than under the assumption of constant concepts of the DEA and the balanced scorecard (BSC)
returns to scale [Charnes 1994]. methods. This approach is then applied to a case study of an
industrial research laboratory that selects from dozens of R&D
Input-oriented models try to find a virtual unit by minimising
projects every year. A combination of DEA and BSC is also used
inputs while maintaining a given level of outputs. In this model,
in the article by [Sadeghani 2013]. The authors state that even
relative technical efficiency is expressed as the ratio between
though BSC and DEA are two different approaches, they
weighted outputs and weighted inputs, while meeting the
complement each other and their combination is therefore
condition that the efficiency rates of all other units are lower
useful. DEA is able to overcome the limitations of the BSC
than or equal to one. A DMU with a relative technical efficiency
method and provides managers with additional useful
ratio equal to one is efficient, a coefficient lower than one
information. On the other hand, BSC provides convenient
identifies the unit as inefficient [Jiang 2011].
inputs and outputs for the DEA model. The DEA method can be
Output-oriented models try to find a virtual unit by maximising used for project quality assessment that is characterised by
outputs while maintaining the level of inputs. In this model, multiple variables and variable returns to scale [Zhang 2006].
relative technical efficiency is expressed as the ratio between Specifically, a DEA CCR model was designed, including
weighted outputs and weighted inputs, while meeting the expansion models for calculating a quality score that serves as
condition that the efficiency rates of all other units are greater the basis for assessment. The authors present a case study in
than or equal to one. A DMU with a relative technical efficiency which they apply the DEA method to assess 10 selected ITECHS
ratio equal to one is efficient, a coefficient greater than one projects and 20 projects from the SourceForge.net portal within
identifies the unit as inefficient [Jiang 2011]. two groups of data using five input/output metrics. The results
show that this approach is efficient in assessing project quality
If a DMU is classified as efficient in the CCR model, it is also
and makes it possible to obtain accurate estimates of future
efficient in the BCC model, but this does not apply the other
improvements. The authors of the article [Xu 2011] have
way round [Cooper 2006]. Depending on the specific type of
provided a different perspective on project performance
model and the relationship between the number of DMUs and
assessment. They develop two performance assessment
the number of inputs and outputs, multiple units may be
processes based on expected and actual performance
classified as efficient. Due to the possibility of further
objectives, while drawing a distinction between two aspects of
classification, super-efficiency models have been proposed in
performance: effectiveness and efficiency. While effectiveness
which the efficient units receive a super-efficiency rate greater
assessment is done through a multi-criteria optimisation
than one (for input-oriented models) or lower than one (for
model, efficiency assessment is based on the DEA model. The
output-oriented models). The best-known super-efficiency
authors conclude that the DEA model provides the relative
models are: the radial super-efficiency model developed by
assessment of project performance and identifies possible ways
Andersen and Petersen, the directional distance function super
to improve inefficient projects. The results of this study can
efficiency model developed by Ray and the SMB super
provide managers with insights in assessing project
efficiency model developed by Tone [Zhang 2017].
performance. The authors of [Shirouyehzad 2011] have also
The DEA method can be used to assess projects ’ relative developed an approach that can help managers efficiently
efficiency and performance using a combination of multiple evaluate the performance of each project relative to the best
inputs and outputs that affect project performance [Yüksel project. The authors use the BCC model, which is applied to 12
2012]. The method can be used to assess the efficiency of DMUs having one input and two outputs. According to the
projects and then to prioritise them, to identify efficient and authors, the proposed methodology – as mentioned in that
inefficient projects, to present reasons for inefficient projects, study – may help managers to quantify project efficiency. The
and to analyse factors that prevented projects from being DEA method can be used as part of an integrated methodology
efficient. Last but not least, the method can be used to assess for the evaluation and priority ranking of new product
the technical and allocation efficiency of the actual project development projects [Hung 2009]. Its authors use fuzzy
teams. Specific applications of the DEA method in publications hierarchical analysis to determine the weights of assessment
of authors focusing on the selected area are addressed in the criteria and the DEA method to analyse efficiency, in order to
next section of this article. identify NPD projects with market potential and high added
2.2 Using Data Envelopment Analysis in project assessment value. The authors of the article [Yang 2010] assess 63 software
projects at a major Canadian bank. The chosen DEA model was
In studying possible approaches to project assessment, the
developed to measure software project efficiency, with a focus
authors focused on researching publications in which the DEA
on factors that affect software productivity. Here, the dummy
method was used for project assessment purposes. Projects
variable was used as the input and two production ratios
were considered to be decision making units. Until recently,
operating with the cost, duration and size of the project were
DEA had been used mainly to study projects within specific
used as the outputs. In the publication [Yang 2015] the DEA
functional areas. In a case study to evaluate the performance of
method is used to assess the operational efficiency of more
engineering design projects, the DEA method was applied in
than a thousand healthcare projects implemented at the
order to compare projects within the engineering department
National Institutes of Health in New York. The authors place the
of Belgian Armed Forces [Farris 2006]. The authors constructed
main emphasis on the study of environmental variables that
a performance index that takes into account project duration as
significantly affect project performance.
an output and also the key input variables that affect the
duration (effort, project staffing, priority, number of officers, Since the authors of this article focused on lean management
and technical complexity). The application of the DEA method and improvement projects, they have studied scientific articles
proved to be convenient in the assessment and subsequent in which the authors used the DEA method in connection with

MM SCIENCE JOURNAL I 2018 I MARCH

2307
tools supporting process improvement such as Total Productive 2.3 Single Minute Exchange of Die
Maintenance (TPM) or Six Sigma. In [Wang 2006], the DEA Single Minute Exchange of Die (SMED) or quick changeover is
method is applied in checking TPM implementation one of the lean production methods used to reduce set up
performance. The objective was to assess the relative efficiency times. Quick changeover makes it possible to significantly
of 53 plants based on three inputs and four outputs. A CCR reduce the machine set-up time during changeover from one
input-optimisation model working with constant returns to type of manufactured product to another, eliminates
scale was used, which was subsequently assessed using wastefulness that is associated with changeover, and ensures
Frontier Analyst. In addition, the authors of the [Jeon 2011] flexible production [Shingo 1985].
study also use DEA to measure the efficiency of TPM
The objective of this method is to reduce set-up times (the time
implementation, but with regard to the overall process of TPM
needed to exchange tools and preparations, i.e. the time
implementation in a three-stage model. Finally, the authors of
between the completion of the last high-quality piece from a
the article [Turanoglu Bekar 2016] propose a new framework
given production batch to the production of the first high-
for assessing the performance of the TPM method using a
quality piece of the next production batch) [Ferguson 2013].
fuzzy-DEA model.
This method was developed by Japanese industrial engineer
The authors of the article [Kumar 2007] used the DEA method Shigeo Shingo, who applied it to help several companies to
to select Six Sigma projects. They identified important inputs significantly reduce machine set-up times. His pioneering work
and outputs associated with the introduction of Six Sigma resulted in an average set-up time reduction of 94% (e.g. from
projects. They gave a hypothetical example and analysed a 90 minutes to less than five minutes). The increasing diversity
dataset of 20 fictitious projects using a DEA tool. In the same of products led to an increase in the number of changeovers
year, several other authors also described potential from one type of manufactured product to another. Each
applications of the DEA method for selecting Six Sigma projects. transition required a new set-up and, in turn, led to losses of
According to [Mawby 2007], the main objective of the DEA valuable production time as a result of increased idle time.
approach to selecting the Six Sigma project portfolio is to Having spent many years working to solve this problem, Shigeo
determine the priorities for each project more objectively than Shingo came up with a method that would reduce the entire
would be the case if most other methods were used. Assessing set-up time to single-digit figures, thus saving useful production
the performance of Six Sigma projects is an important issue for time that would otherwise be lost during machine set-up
companies that apply Six Sigma projects [Yüksel 2012]. Yüksel [Mukherjee 2006].
uses an input-oriented DEA model to assess the performance of
Set-up operations are divided into internal set-up operations,
Six Sigma projects. The case study included only 5 projects and
which are only performed when the machine is switched off,
the author identified two inputs (hours worked and costs per
and external set-up operations, which can be performed even
project) and three outputs (financial gains, increase in sigma
when the machine is running. The basic solution for successful
level, and increase in customer satisfaction). The authors of the
SMED implementation is to reduce both set-up times and to
article [Meza 2013] aim to assess the performance of individual
transfer elements from internal operations to external
Lean Six Sigma projects and develop recommendations for
operations, thus reducing production equipment downtime
strategies to improve operational efficiency using the DEA
[Wang 2011]. After transferring internal operations to external
method. An important part consisted in a survey that provided
ones, the required internal set-up time can be reduced by 30 %
a basis for identifying the critical factors for project success,
to 50 % [Shingo 1985].
which were subsequently used as inputs of the DEA model. A
different approach can be found in [Alinezhad 2013], where the The SMED method makes it possible to significantly reduce
DEA method was applied to interval data in order to select changeover times. Quick changeover makes it possible to do
high-priority Six Sigma projects with maximum financial changeovers more frequently and reduce the size of
benefits to the organization. In order to obtain a full ranking of economically viable production batches. This leads to a
projects, a dummy project with maximum of inputs and reduction in inventory and, consequently, to better quality
minimum of outputs was used. The authors of the article control and waste reduction. The SMED method makes it
[Yousefi 2014] used Linear Discriminant Analysis to verify that possible to reduce the production lead time and deliver
the DEA model proposed by them was suitable for selecting Six products to the customer in a timely manner. Other effects of
Sigma projects, and presented a case study from the electricity SMED implementation include increased productivity,
distribution industry. The most important step in reducing the elimination of setup errors, improved quality, and increased
risk of Six Sigma projects’ failure is to successfully select those safety [Wang 2011].
with the most benefits and fewest risks. According to the
authors of the article [Arafah 2015], there are many different
3 METHODOLOGY
formulations of the DEA model that can influence both the
selection process and the final choice of the project. The This study was implemented in a fictitious company operating
success of a Six Sigma initiative is then affected by successful in the automotive industry. 20 hypothetical SMED projects
project selection at the beginning. These authors apply nine were selected for Data Envelopment Analysis. It involves the
different DEA formulations to several case studies and conclude introduction of the SMED tool on some of the 20
that different DEA formulations result in the selection of pressing/welding machines and the company is deciding which
different projects. Finally, the authors of the article [Bazrkar of them should be selected for implementation. However, the
2017] focus – in their study – on identifying priorities and company’s budget does not allow the implementation of all
selecting the best Lean Six Sigma project using the cross- projects; the company wants to implement only those projects
efficiency model within the DEA method. that are found to be efficient according to the presented DEA
method and, at the same time, that represent the highest
Process improvement is associated, among other things, with potential and value added.
reducing process duration or, if relevant, the duration of sub-
Below, the definition of the set of DMUs is followed first by the
activities. The following section briefly describes the SMED tool,
identification of possible input and output variables and then
which is used to reduce the machine set-up time.

MM SCIENCE JOURNAL I 2018 I MARCH

2308
by the construction of an input-oriented CCR-I model. After Tab. 1 and 2 show the inputs and outputs that were used for all
that, the model is applied to data to give an example of its use 20 hypothetical SMED projects.
for relative efficiency assessment and subsequently the
selection of one or more SMED projects for implementation. Project I1 (CZK) I2 (days) I4 (CZK)
3.1 Definition of Inputs and Outputs parameters
Project 1 18,876 64 18,450
Project selection criteria can be divided into three categories:
Project 2 14,157 81 15,450
bottom-line criteria (customer impact, impact on business
strategy, impact on core competencies, financial impact); Project 3 21,054 51 7,800
feasibility criteria (required resources, available expertise, Project 4 14,520 50 29,850
implementation complexity, probability of success), and
Project 5 35,816 82 2,300
organisational impact criteria (educational benefits, benefits for
learning and growth, cross-functional benefits) [Misra 2008]. Project 6 21,054 65 9,850
Project 7 18,876 78 16,980
Below, attention is given to defining potential inputs and
outputs. The measures to be minimised are considered as Project 8 19,360 50 8,750
inputs (I), while the measures to be maximised are considered Project 9 19,360 56 14,500
as outputs (O).
Project 10 14,157 78 18,950
The following inputs can be used: (I1) project costs (training Project 11 35,816 65 7,560
and upgrading-training), (I2) project preparation time, (I3) cost
Project 12 28,072 57 1,900
of modifications and changes at the workplace (e.g.
construction work), (I4) cost of resources helping reduce time Project 13 28,072 51 12,650
required for changeover to another type of product (quick- Project 14 26,862 82 5,500
release jigs, consumables, lubricants, detergents).
Project 15 18,876 82 17,500
Furthermore, the following outputs can be tracked in assessing Project 16 14,520 56 17,980
SMED projects: (O1) reduction in external set-up time, (O2)
Project 17 21,054 57 6,450
reduction in internal set-up time (time spent searching, waiting,
walking, setting-up), (O3) reduction in set-up errors, (O4) Project 18 14,520 64 22,580
reduction in production lead time, (O5) reduction in Project 19 35,816 79 13,550
unnecessary movements, (O6) increase in work safety, (O7)
Project 20 26,862 79 10,250
increase in machine utilisation rate, (O8) increased
productivity, (O9) reduction in inventory of spare parts and Table 1. Data used as inputs in DEA modelling
accessories. Another important indicator that can be tracked is
(O10) changeover time, i.e. the time that is needed during
changeover from manufacturing one product to another Project O7 (%) O10 (min) O11 (%)
product. This indicator is the sum of the times of the following Project 1 5 25 7
four activities: preparation time, tool change time, set-up time,
and inspection time. Indicator (O11) First Time Through (FTT) Project 2 6 40 5
makes it possible to track production process quality. It can be Project 3 9 25 9
calculated as the ratio of the number of good parts to the total Project 4 7 50 4
number of parts. Another important indicator is (O12) OEE,
Project 5 2 20 15
which takes into account three sub-indicators – (Availability of
equipment, Performance of equipment and Quality of Project 6 4 30 7
production on the equipment. Project 7 9 45 8
It is known that too many inputs and outputs as compared with Project 8 8 25 10
the number of DMUs may adversely affect the discriminatory Project 9 13 35 9
power of the selected DEA model [Zizka 2017]. There are
Project 10 14 50 4
various rules, one of the best-known rules states that the
number of DMUs should be at least twice or even three times Project 11 14 15 11
the number of inputs and outputs [Raab 2002]. In this paper, 3 Project 12 5 12 12
input and 3 output variables have been specified, while there Project 13 6 15 13
are 20 DMUs. Since most inputs and outputs have a
probabilistic nature, their expected values have been used. The Project 14 3 10 10
inputs and outputs used are expressed in different units of Project 15 12 30 9
measurement. Project 16 15 48 5
Based on completed correlation and regression analysis, the Project 17 11 15 10
following inputs have been used for measuring technical Project 18 12 55 6
efficiency: (I1) project costs (CZK), (I2) project preparation time
(days), and (I4) cost of resources helping reduce time required Project 19 4 30 12
for changeover to another type of product (CZK). The following Project 20 2 30 10
outputs have been chosen: (O7) increase in machine utilisation
rate (%), (O10) reduction in changeover time (minutes), and Table 2. Data used as outputs in DEA modelling
(O11) FTT (%).

MM SCIENCE JOURNAL I 2018 I MARCH

2309
3.2 Construction of the DEA model Project CCR-I efficiency Efficiency
The paper assumes that projects have constant returns to scale scores (OTE) ranking
and that inputs are easier to control than outputs. That is why
Project 1 0.746795 8.
an input-oriented CCR-I model operating on the assumption of
constant returns to scale is used for selecting efficient DMUs. Project 2 0.960437 3.
This means that organisations are able to linearly transform Project 3 1.000000 1.
inputs into outputs without increasing or reducing efficiency. Project 4 1.000000 1.
Input Oriented CCR-I model Project 5 1.000000 1.
Suppose, that we have n DMUs where each DMUj, j = 1, 2, ..., n,
Project 6 0.948098 4.
produces the same s outputs, Yrj (r = 1, 2, ..., s), using the same
m inputs, Xij (i = 1, 2, ..., m). The efficiency of a specific DMUq Project 7 0.987953 2.
can be evaluated by the CCR model of DEA. Its dual form can be Project 8 1.000000 1.
formulated as: Project 9 1.000000 1.
Project 10 1.000000 1.
 m s

E0  min .     si   s r  Project 11 1.000000 1.
 i 1 r 1  Project 12 1.000000 1.
n Project 13 1.000000 1.
s.t.  j X ij si  X i 0 , i  1,..., m, Project 14 0.810633 6.
j 1
Project 15 1.000000 1.
n

 Y
j 1
j rj

 s  Yr 0 , r  1,..., s,
i (1) Project 16
Project 17
1.000000
1.000000
1.
1.

 j , s , s  0, j  1,..., n, i  1,..., m, r  1,..., s.


i

i
Project 18 1.000000 1.
Project 19 0.766052 7.
 unrestricted in sign. Project 20 0.881317 5.
where λj, j = 1, 2, …, n are weights of all DMUs, s-i,
i = 1, 2, …, m Table 3. Results of CCR-I model
and s+r, r = 1, 2, …, s are slack/surplus variables, ε > 0 is non-
Archimedean element defined to be smaller than any positive In a situation where resources in the organisation are limited, it
real number, θ is the efficiency score of the DMUq that is desirable to be able to appropriately allocate them to the
expresses the reduction rate of inputs in order this unit reaches most beneficial projects, which may be the best of the efficient
the efficient frontier. projects. As mentioned above, the conventional CCR-I model
does not rank efficient projects in any way. Therefore, the
For the further classification of efficient projects, the CCR-I super-efficiency model described in Section 3.2 will be used,
model was supplemented with Andersen and Petersen super- which makes it possible to rank efficient projects according to
efficiency model. Therefore, it will be possible to create a their super-efficiency score values. In this concept efficient
ranking of all projects from efficient to inefficient. Its projects acquire a super-efficiency score greater than 1. The
formulation is very close to the standard formulation of the higher the value of the super-efficiency score, the better the
CCR-I model. Only difference is that the weights of the DMUq, project evaluation.
i.e. λq, are set to zero:

q  0 (2)
Project CCR-I super-
efficiency
Super-efficiency
ranking
scores
This causes that the DMUq is removed from the set of units and
the efficient frontier changes its shape after this removal. Project 1 0.746795 20.
4 RESULTS Project 2 0.960437 15.
The next step is to calculate the relative technical efficiency of Project 3 1.060718 11.
each project in order to identify efficient projects. The Project 4 1.163636 7.
calculation was made in the DEA Solver for MS Excel 2010
Project 5 1.376812 2.
environment.
Project 6 0.948098 16.
The CCR-I model assumes that the company operates at the
optimal scale, i.e. under conditions of constant returns to scale Project 7 0.987953 14.
(CRS). The technical efficiency score determined using the CCR Project 8 1.130704 8.
model is called overall technical efficiency (OTE). Project with Project 9 1.082840 10.
OTE equal to 1 is efficient, whereas OTE lower than 1 indicates
Project 10 1.021522 12.
an inefficient project. Subsequently, the ranking of the projects
was compiled based on the OTE values. Project 11 1.109562 9.

Tab. 3 shows that projects 3, 4, 5, 8, 9, 10, 11, 12, 13, 15, 16, 17 Project 12 1.806999 1.
and 18 are efficient and, in turn, eligible for implementation in Project 13 1.218244 5.
the process. Project 14 0.810633 18.
Project 15 1.007958 13.

MM SCIENCE JOURNAL I 2018 I MARCH

2310
 Project 16 1.313843 3. group if the budget of the resources defined to cover project
Project 17 1.261104 4. costs is limited.

Project 18 1.175472 6. There is no doubt that the company can achieve many benefits
if SMED is properly implemented and standardized. But the first
Project 19 0.766052 19.
and essential step is to select the right project to be
Project 20 0.881317 17. implemented. The authors of this article introduced the
approach that can make this selection easier in practice.
Table 4. Results of CCR-I super-efficiency model
Depending on the availability of data, there are several other
Tab. 4 lists the projects in new and more precise order
ways to extend DEA application. In further research, we would
according to their super-efficiency score values. Because the
like to apply the DEA method on real data in other companies.
efficient project 12 is the first one, it should have priority for
Besides the SMED tool, we consider to evaluate the
implementation, and the implementation of other projects
effectiveness of another lean management tool.
would depend on the availability of resources in the
organisation. To shorten the set-up time, the priority is given to
measures that do not require additional resources, but which REFERENCES
offer significant potential for improvement. [Alinezhad 2013] Alinezhad, A. et al. Selecting Six Sigma
In the company, for example, could be set an amount of CZK projects with interval data envelopment analysis. International
400,000 as the maximum of available resources of project costs Journal of Quality and Innovation, January 2013, Vol. 2, No. 2,
(staff training and upgrading-training) and the cost of resources pp 138-147. ISSN 1756-6975
helping reduce time required for changeover to another type of [Arafah 2015] Arafah, M. Selecting the Six Sigma Project: A
product. The sum of these costs for all 13 effective projects is Multi Data Envelopment Analysis Unified Scoring Framework.
CZK 453,967. Taking into account the maximum amount of the American Journal of Operations Research, May 2015, Vol. 5,
budget, it is obvious that implementation of two effective No. 3, pp 129-150. ISSN 2160-8830
projects (namely project 15 and project 10) would be [Bazrkar 2017] Bazrkar, A. and Iranzadeh, S. Prioritization of
abandoned. Consequently, 11 efficient SMED projects would be Lean Six Sigma Improvement Projects using Data Envelopment
selected and the sum of the monitored cost components would Analysis Cross Efficiency Model. Romanian Society for Quality
be CZK 384,484. Assurance, April 2017, Vol. 18, No. 157, pp 72-76. ISSN 1582-
2559
[Bogetoft 2011] Bogetoft, P. and Otto, L. Benchmarking with
5 CONCLUSIONS DEA, SFA, and R. New York: Springer, 2011. ISBN 978-144-1979-
The present article deals with the use of the DEA method in 605
addressing a major issue that concerns most manufacturing [Charnes 1994] Charnes, A. et al. Data envelopment analysis:
businesses, i.e. in assessment and selecting SMED projects. theory, methodology, and application. Boston: Kluwer
Academic Publishers, 1994. ISBN 978-079-2394-808
First, cross-sectional resource search was conducted focusing
[Cook 2005] Cook, W. D. and Zhu, J. Modeling performance
on one of the methods of multi-criteria evaluation of variants,
measurement: applications and implementation issues in DEA.
the DEA method. The use of DEA method in the assessment and
New York: Springer, 2005. ISBN 03-872-4137-X
selection of projects was theoretically examined. Due to the
[Cooper 2006] Cooper, W. W. et al. Introduction to data
focus of this article on lean management and process
envelopment analysis and its uses: with DEA-solver software
improvement projects, the use of the DEA method was
and references. New York: Springer, 2006. ISBN 978-038-7285-
followed in connection with tools to support process
801
improvement.
[Eilat 2008] Eilat, H. et al. R&D project evaluation: An
A hypothetical case study then deals with a selected tool to integrated DEA and balanced scorecard approach. Omega,
support process improvement, the SMED tool, which is used to October 2008, Vol. 36, No. 5, pp 895-912. ISSN 0305-0483
reduce the machine set-up time. In the case study, the DEA [Farris 2006] Farris, J. A. et al. Evaluating the Relative
method was implemented in a fictitious company operating in Performance of Engineering Design Projects: A Case Study
the automotive industry. The aim was to select from a set of 20 Using Data Envelopment Analysis. IEEE Transactions on
hypothetical SMED projects the effective projects that would Engineering Management, August 2006, Vol. 53, No. 3, pp 471-
be appropriate to be implemented, and to create their order 482. ISSN 0018-9391
from the most efficient to the least efficient. [Ferguson 2013] Ferguson, D. L. Removing the barriers to
efficient manufacturing: real-world applications of lean
The authors dealt with the identification of potential input and productivity. Boca Raton: CRC Press, 2013. ISBN 978-146-6555-
output variables of the DEA model that can be used to assess 518
SMED projects. Of these, 3 inputs and 3 outputs were selected [Gupta 2014] Gupta, P. K. and Garg S. DEA Modeling for
for the case study. An input-oriented CCR-I model operating on Efficiency Optimization of Indian Banks with Negative Data Sets.
the assumption of constant returns to scale was used for In: D. Huisman, I. Louwerse and A. P. M. Wagelmans, eds.
selecting efficient projects. For the purpose of the further Proceedings of the International Conference on Operations
classification of efficient projects, the above-mentioned Research, Rotterdam, 3-6 September, 2013. Switzerland:
efficiency model was supplemented with Andersen and Springer International Publishing, pp 169-176. ISBN 978-331-
Petersen super-efficiency model. Applying both models to 9070-001
hypothetical data, 13 SMED projects were selected, which [Hung 2009] Hung L. C. et al. Using DEA Approach to Develop
could be described as effective and recommended for the Evaluation and Priority Ranking Methodology of NPD
implementation. All projects were then ranked from the most Projects. In: S. Y. Chou, A. Trappey, J. Pokojski and S. Smith, eds.
effective to the least effective project. This allows the selection Proceedings of the Global Perspective for Competitive
of a few of the best-rated projects from the effective project

MM SCIENCE JOURNAL I 2018 I MARCH

2311
Enterprise, Economy and Ecology, Taiwan, 20-24 July, 2009. Envelopment Analysis. In: S. I. Ao, L. Gelman, D. W. L. Hukins, A.
London: Springer, pp 159-166. ISBN 978-1-84882-761-5 Hunter and A. M. Korsunsky, eds. Proceedings of the World
[Jeon 2011] Jeon, J. et al. Measuring efficiency of total Congress on Engineering, London, 29 June-1 July, 2016. London:
productive maintenance (TPM): a three-stage data Newswood Limited, pp 135-140. ISBN 978-988-19253-0-5
envelopment analysis (DEA) approach. Total Quality [Wang 2006] Wang, F. Evaluating the efficiency of
Management & Business Excellence, September 2011, Vol. 22, implementing total productive maintenance. Total Quality
No. 8, pp 911-924. ISSN 1478-3363 Management & Business Excellence, June 2006, Vol. 17, No. 5,
[Jiang 2011] Jiang, X. et al. The Online Advertising Types of pp 655-667. ISSN 1478-3363
Choices Based on DEA Method. In: Y. Shi, S. Wang, G. Kou and J. [Wang 2011] Wang, J. X. Lean manufacturing: business bottom-
Wallenius, eds. New state of MCDM in the 21st century: line based. Boca Raton: CRC Press, 2011. ISBN 978-142-0086-
selected papers of the 20th International Conference on 027
Multiple Criteria Decision Making, Chengdu, 21-26 June, 2009. [Xu 2011] Xu, Y. and Yeh, C. Evaluating the performance
New York: Springer, pp 157-166. ISBN 978-364-2196-942 effectiveness and efficiency of projects. In: W. Xie, ed.
[Kumar 2007] Kumar, U. D. et al. Six sigma project selection Proceedings of the 6th IEEE Conference on Industrial
using data envelopment analysis. The TQM Magazine, August Electronics and Applications (ICIEA), Beijing, 21-23 June, 2011.
2007, Vol. 19, No. 5, pp 419-441. ISSN 0954-478X New Jersey: IEEE, pp 1090-1094. ISBN 978-1-4244-8754-7
[Linton 2007] Linton, J. D. et al. An extension to a DEA support [Yang 2010] Yang, Z. and Paradi, J. C. A DEA evaluation of
system used for assessing R&D projects. R&D Management, software project efficiency. Proceedings of the IEEE
January 2007, Vol. 37, No. 1, pp 29-36. ISSN 1467-9310 International Conference on Industrial Engineering and
[Mantri 2008] Mantri, J. K. Research methodology on data Engineering Management, Hong Kong, 8-11 December, 2009.
envelopment analysis (DEA). Boca Raton: Universal-Publishers, New York: IEEE, pp 1723-1727. ISSN 2157-3611
2008. ISBN 978-159-9429-502 [Yang 2015] Yang, Z. et al. DEA evaluation of health project
[Mawby 2007] Mawby, W. D. Project portfolio selection for Six performance. Proceedings of the IEEE International Conference
Sigma. Milwaukee: ASQ Quality Press, 2007. ISBN 978-087- on Progress in Informatics and Computing (PIC), Nanjing, 18-20
3897-075 December, 2015. New York: IEEE. ISBN 978-1-4673-9088-0
[Meza 2013] Meza, D. and Jeong, K. Measuring efficiency of [Yao 2010] Yao, S. et al. Performance of the Chinese insurance
lean six sigma project implementation using data envelopment industry under economic reforms. Northampton, MA: Edward
analysis at Nasa. Journal of Industrial Engineering and Elgar, 2010. ISBN 978-184-7203-816
Management, January 2013, Vol. 6, No. 2, pp 401-422. ISSN [Yousefi 2014] Yousefi, A. and Aqamohammadi, A. R. A new
2013-8423 DEA model for six sigma project selecting: Case study on
[Misra 2008] Misra, K. B. Handbook of performability Esfahan Province Electricity Distribution Co (EPEDC). In: XX ,eds.
engineering. London: Springer, 2008. ISBN 978-184-8001-305 Proceedings of the IEEE International Conference on Industrial
[Mukherjee 2006] Mukherjee, P. N. Total quality management. Engineering and Engineering Management (IEEM), Bandar
New Delhi: Prentice-Hall of India, 2006. ISBN 81-203-3056-0 Sunway, 9-12 December, 2014. New Jersey: IEEE, pp 627 - 631.
[Raab 2002] Raab, R. L. and Richard, W. L. Identifying Subareas ISBN 978-1-4799-6410-9
That Comprise A Greater Metropolitan Area: The Criterion of [Yüksel 2012] Yüksel, H. Evaluation of the Success of Six Sigma
County Relative Efficiency. Journal of Regional Science, Projects by Data Envelopment Analysis. International Journal of
December 2002, Vol. 42, No. 3, pp 579-594. ISSN 0022-4146 Business and Management, July 2012, Vol. 7, No. 13, pp 75-84.
[Ramanathan 2003] Ramanathan, R. An introduction to data ISSN 1833-3850
envelopment analysis: a tool for performance measurement. [Zhang 2006] Zhang, S. et al. Evaluation of Project Quality: A
New Delhi: Sage Publication, 2003. ISBN 07-619-9760-1 DEA-Based Approach. In: Q. Wang, D. Pfahl, D. M. Raffo and P.
[Ray 2004] Ray, S. C. Data envelopment analysis: theory and Wernick, eds. Proceedings of the Software Process Change,
techniques for economics and operations research. New York: Shanghai, 20-21 May, 2006. Berlin: Springer, pp 88-96. ISBN
Cambridge University Press, 2004. ISBN 05-218-0256-3 978-3-540-34199-4
[Sadeghani 2013] Sadeghani, M. et al. Analysis of the Efficiency [Zhang 2017] Zhang, Z. Environmental data analysis: methods
R&D Projects base on BSC-DEA Approach with Restrictions on and applications. Boston: de Gruyter, 2017. ISBN 978-311-0424-
Weight of Inputs and Outputs. International Journal of 911
Economy, Management and Social Sciences, October 2013, Vol. [Zizka 2017] Zizka, M. An Assessment of the Efficiency and
2, No. 10, pp 775-779. ISSN 2306-7276 Effectiveness of the Services of Urban Transport Operators in
[Sherman 2006] Sherman, H. D. and Zhu, J. Service productivity the Czech Republic. Transformations in Business and
management: improving service performance using data Economics, January 2017, Vol. 16, No. 1, pp 134-152. ISSN
envelopment analysis (DEA). New York: Springer, 2006. ISBN 1648-4460
978-0-387-33211-6
[Shingo 1985] Shingo, S. A revolution in manufacturing: the
SMED system. Stamford: Productivity Press, 1985. ISBN 09-152- CONTACTS:
9903-8 Ing. Eva Stichhauerova, Ph.D.
[Shirouyehzad 2011] Shirouyehzad, H. and Dabestani, R. Safety Ing. Natalie Pelloneova
Approach for Measuring Efficiency of Projects, Using Data
Envelopment Analysis. In: M. Zhang, ed. Proceedings of the Technical University of Liberec, Faculty of Economics,
Second International Conference on Construction and Project Department of Business Administration and Management
Management, Singapore, 16-18 September, 2011. Singapore: Studentska 2, Liberec, 461 17, Czech Republic
IACSIT Press, pp 59-63. ISBN 978-981-08-9176-3 Tel.: +420 48535 2363, [email protected]
[Turanoglu Bekar 2016] Turanoglu Bekar, E. and Kahraman, C. Tel.: +420 48535 2353, [email protected]
Measuring Efficiency of Total Productive Maintenance (TPM) www.tul.cz
with Newly Developed Performance Measures using Fuzzy Data

MM SCIENCE JOURNAL I 2018 I MARCH

2312