1
New product Go/NoGo evaluation at the front end-a fuzzy linguistic
approach
*Ching-Torng Lin and Chen-Tung Chen
Department of Information Management, Da-Yeh University, Changhua, Taiwan,
R.O.C.
Abstract
The screening of a new product concept is perhaps the most critical activity in new
product development (NPD), yet such screening is often not performed well. .
Limited by both the nature and the timing of NPD, new product screening is associated
with uncertainty, imprecision and complexity. This paper discusses an actual illustration
of a new product screening analysis in the development of a new machining center.
Because subjective considerations such as competitive advantage in the market ,
product superiority, technological appropriateness, and product risk were relevant to the
Go/NoGo decision, a fuzzy logic approach is adopted. In this approach
measurements are described subjectively by linguistic terms, while success attributes
are weighted by their corresponding importance using fuzzy values. . The fuzzy
logic-based screening model can efficiently aid managers in dealing with ambiguity and
complexity in product screening decisions.
Keywords: New product screening; New product Go/NoGo decision; New product
front-end decision; Linguistic multi-criteria decision; Fuzzy number.
*Corresponding author:
Address: 112 Shan-Jiau Rd., Da-Tsuen, Changhua, Taiwan 51505, R.O.C.
Tel: 886-4-851-1888 ext. 3133;
Fax: 886-4-851-1500;
E-mail: [email protected]
2
I. INTRODUCTION
Currently many companies are facing increasing competition created by
technological innovations, changing market environments and changing customer
demands. Companies have realized that accelerated new product development (NPD) is
crucial for their survival and winning of competitive battles. Successful NPD can
provide increased sales, profits, and competitive advantage for most companies; yet it is
a complex process and involves business risk. Although NPD requires substantial
monetary and nonmonetary commitments, the costs of a possible failure are higher[1].
Previous research has concentrated on developing models addressing the different
stages of the NPD process to help managers improve their decision making [2]. Even
with an improvement NPD process, competition and emerging new technologies can
still limit the NPD success rate to no more than 59%, and it still requires 6.6 ideas to
generate a successful product [3], the same level as 10 years ago.
The screening of a new product is perhaps the most critical step in the NPD
process. In a study of Canadian manufacturing firms, Cooper and Kleinschmidt [4]
found that initial screening has the highest correlation with new product performance
when compared with a dozen other NPD activities. From a managerial viewpoint,
terminating an inferior product prior to commercialization results in large cost
savings, because costs generally increase dramatically as NPD projects move toward
commercialization. These sunk costs frequently influence decision-makers’ future
Go/NoGo decisions on new products [5]. Several studies have found that it is difficult
for managers to terminate NPD projects once they have begun [6], [7]. In addition, in
many situations, a failing NPD project may be more costly than a successful project [8].
In order to prevent an organization from misallocating its resources in developing a
failing project, researchers [1], [3], [5] maintain that any inferior new product projects
3
should be eliminated at the front end before they lead to a significant investment.
New-product screening decisions are associated with complexity, uncertainty and
imprecision t for the following reasons [9], [10]:
At the time of the decision, usually only uncertain and incomplete information
is available.
The competitive environment is marked by uncertainty and rapid changes in
technologies and markets.
The criteria for a product’s Go/NoGo decision are not always quantifiable or
comparable; criteria may directly conflict or interact with one another other.
Multiple functional groups, each with a different perspective, may be involved
in the evaluation decision.
To assist managers in making better screening decisions, numerous decision tools
have been developed with the hope that managers could make better decisions in an
uncertain environment . However, traditional project selection techniques tend to
utilize quantitative tools, such as mathematical programming, economic models, etc.
which have both practical and theoretical limitations [11], [12]. Amajor obstacle is the
amount of data required: information on the size of the target market; projected
financial returns; resource needs; timing of decisions and probabilities for the
completion and success of the product. Much of this information simply is not available,
and when it is, its reliability can be suspect. Further, all these models tend to ignore
human behavior in the organizational setting; managers may be unable to handle
multiple and interrelated criteria. Uncertainty, complexity and scarce or unreliable
information become a threat to the use of traditional quantitative techniques.
Since humans have the capability of understanding and analyzing obscure or
imprecise events and factors which are not easily incorporated into existing analytical
methods [13], experts' judgments are vital elements in decisions involving uncertainty
4
and ambiguity [11], [14]. To overcome the limitations of quantitative methods, several
qualitative or heuristic approaches, e.g., analogies, expert opinions, intentions, scenario
analyses and information acceleration, focus groups, and decision analysis (see review
in [2]) have been proposed. According to Rangaswamy and Lilien [15], management
science techniques for screening new product ideas can be broadly grouped into three
categories: (1) multicriteria decision making techniques, (2) the Analytic Hierarchy
Process (AHP), and (3) screening regression models. Although, these models can
overcome some the limitations of quantitative methods, they may not adequately
address the ambiguity and multiplicity of possible considerations in the product
screening decision, resulting in an evaluation that is economically sound but
dysfunctional for the organization. The pros and cons of these methods are listed in
Table I.
According to a study conducted by Karwowski and Mital [22], when a situation is
characterized by either lack of evidence or the inability of experts to make a significant
measurement of the possibility of an event, the experts simply adjudge that the score of
a given event is “low,” “high,” or “fairly high.” In other words, it is difficult to assign a
crisp value to a subjective judgement since the data/information is imprecise and
ambiguous. Linguistic terms may alsocontain ambiguity and multiplicity of meanings.
However, the lack of a better approach for interpreting the semantics of these subjective
judgements makes it unrealistic in estimating the success-possibility of an NPD????.
Fuzzy logic is a useful tool fo capturing the ambiguity and multiplicity of meanings of
the linguistic expression. That is why we propose to use fuzzy logic in the new product
Go/NoGo decision.
II. FUZZY LOGIC AND APPLICATIONS IN DECISION MAKING
5
A fuzzy set can be defined mathematically by assigning a value to each possible
member in a universe representing its grade of membership. Membership in the fuzzy
set to a greater or lesser degree is indicated by a larger or smaller membership grade.
Fuzzy set methods allow uncertain and imprecise systems of the real world to be
captured through the use of linguistic terms so that computers can emulate human
thought processes. Thus fuzzy logic is a very powerful tool that can deal with decisions
involving complex, ambiguous and vague phenomena that can only be assessed by
linguistic values rather than numerical terms. Fuzzy logic enables one to effectively and
efficiently quantify imprecise information, perform reasoning processes and make
decisions based on vague and incomplete data [11]. Roussel et al. [31] contend that
the experts can manage the risk when it is know, but in uncertain situations when
available information is scarce or unreliable or when target objectives and goals are not
clearly defined, managers often react very poorly. Fuzzy logic, by making no global
assumptions about the independence, exhaustiveness, or exclusiveness of underlying
evidence, tolerates a blurred boundary in definitions [11]. Thus, fuzzy logic brings hope
of incorporating qualitative factors into decision-making,
Fuzzy logic is currently being used extensively in many industrial applications
such as water treatment, traveling time reduction, subway systems, washing machines,
vacuum cleaners, rice cookers, and flight control of aircraft, to name just a few [23].
Fuzzy logic has also been applied to managerial decision making as well. For example,
it has been used in muliti-attribute decision-making situations to select information
system projects [11], [24], and iron-making technology [25]. Ben Ghalia et al. [26] used
fuzzy logic inference for estimating hotel room demand by eliciting knowledge from
the hotel’s’ managers and building fuzzy IF-THEN rules. Since the fuzzy weighted
average approach produces a more informative result, Kao and Liu [27] used this
6
technique to devise a competitiveness index for manufacturing firms based on their use
of automation technology and manufacturing management practices. Lin [28] devised a
fuzzy-possible-success-rating for evaluating whether to bid or not bid on a project
based on the resources, reputation, and mission of the company; the probability of
project go-ahead, and the risk and competition involved the project. Chen and Chiou
[29] devised a fuzzy credit rating for commercial loans. Hui et al. [30] captured the
knowledge of experienced supervisors to create a fuzzy rule-based system for balance
control of assembly lines in apparel manufacturing.
As mentioned previously, the new product screening evaluation processis associated
with uncertainty and complexity. Managers must make a decision by considering
product attributes which may have non-numerical values. They must integrate all
attributes within the evaluation decision, none of which may exactly satisfy the firms’
ideal. Conventional "crisp" evaluation approaches can not handle such decisions
suitably or effectively. Since humans have the capability of understanding and
analyzing obscure or imprecise events which are not easily incorporated into existing
analytical methods; the new productscreening decision is made primarily on the basis
of opinions of experts. Experts have found it easier to express their measurements in
linguistic terms. Linguistic terms usually have meanings which are vague. One way to
capture the meanings of linguistic terms is to use the fuzzy logic approach to associate
each linguistic term with a possibility distribution [32].Using the concepts of
multicriteria decision making and fuzzy logic, we devised a
fuzzy-possible-success-ratin for a newproduct Go/NoGo decision. The fuzzy logic
screening model [FLSM] can efficiently aid managers dealing with ambiguity and
complexity in achieving relatively realistic and informative results in the evaluation
process.
7
III. The FLSM PROCESS
The evaluation framework of FLSM, shown in Figure 1, is composed of three main
parts. The first characterized the context of the new-product development scenario.
Relevant factors include changes in the business environment, the company’s
strategies as well as managerial goals, and the company’s competencies and resources.
These facts/factors affect the company’s competitive positioning and competitive
advantage in the market. This is the basis for determining the relevant factors in new
product screening. The second part of the framework analyzes and synthesizes the
criteria to obtain fuzzy merit importance indices for each criterion and a fuzzy
possible-success rating for a new product. The third part matches the fuzzy
possible-success rating with an appropriate linguistic term for decision-making and
ranks the fuzzy merit-importance indices to identify major adverse factors so that
managers may proactively implement appropriate preventive measures.
A stepwise description of the implementation of the evaluation framework is
given below:
1. Form a committee of decision-makers and collect project-related data.
2. Select criteria for decision making.
3. Define linguistic variables as well as associated membership functions for
assessing the merit ratings and the importance weights of the selected criteria.
4. Assess the criteria rating and weight using linguistic terms.
5. Translate the linguistic ratings and weights into fuzzy numbers.
6. Aggregate fuzzy numbers to obtain fuzzy merit-importance indices of selected
criteria and a fuzzy-possible-success-rating for the new product development
project.
7. Translate the fuzzy-possible-success-rating into an appropriate linguistic term for
8
recommending a Go/NoGo decision.
8. Rank fuzzy merit-importance indices of criterion to identify the primary adverse
factors.
IV. CASE STUDY: GO/NO GO DECISION FOR A NEW MACHINING CENTER
DEVELOPMENT
In this section the development of a new machining center at the Taiwan Victory
(TV) Company is described to illustrate the details of the FLSM and demonstrate how
it can be used in new product screening. It is generally recognized that every firm has
its own set of criteria and evaluation levels in new product screening [20]. Our attempt
here is to present a generalized model based on past studies that can then be modified
or extended for use in a specific situation or company.
A. Subject of Case Study
The model was developed and validated with input from the TV Company, an
internationally renowned machine-tool company, particularly known for CNC lathes.
Its products include conventional lathes, high-precision tools, and machining centers.
To meet an increasingly competitive environment in the machining market, TV has
decided to expand its product line to include large-size horizontal machining centers,
automated flexible manufacturing cells (FMC), and integrated flexible manufacturing
systems (FMS) to supply a global market.
To compete in the 21st century, TV realized that the capability to rapidly develop
new products or improve existing products that users want and will continue to
purchase was crucial for its survival. In tracking product performance and customer
9
needs over time using perceptual mapping and conjoint analysis, the CEO was
convinced that advanced tool-changing and automatic-gauging machining center
systems were desired by machining-tool users. To capture this potential market, the
new product TVcenter-HX, a next generation platform FMS representing a new system
solution for machining-tool users, was proposed. The proposed TVcenter-HX had three
essential characteristics: (1) core performance capabilities that match primary customer
needs, (2) ability to support an entire product/process generation, and (3) linkages to
previous and subsequent generations. TV desired a system architecture to facilitate the
addition of other features or the removal of existing features in order to tailor
derivative products for special niche markets.
B. New-Product Development Screening-Concept Model
TV’s CEO mandated that all new product proposals would be thoroughly analyzed
and evaluated before undergoing full-scale development. In order to determine the
appropriate product and characteristics to be developed, and pursuant with previous
studies TV revised its model for new product screening, which had last been revised in
1993 when the company set up an ISO-9001 compliant system. The model based on
previous studies [33], [34], determined the appropriate product characteristics to be
developed. The model, illustrated in Figure 2, shows the linkage between new
product screening goals and successful new product development..
C. Application the FLSM to the TVcenter-HX Project
On the basis of the procedures of FLSM a decision to launch the TVcenter-HX was
reached. The deliberations over whether to start full-scale development are summarized
10
below:
1) Form a committee of decision-makers and collect project-related data: For
evaluating the TVcenter-HX, a screening committee composed of four
experts/senior-managers from marketing, technology, operations, and finance was
organized and led by the CEO. Each of these members brought particular needs and
desires into the decision which had to be reconciled into a consensus since all parties
would contribute to the success or failure of the decision. The next step was to collect
as wide range of information as possible concerning the TVcenter-HX project.
As mentioned previously, the company had used perceptual mapping to understand
the current market conditions and used conjoint analysis to identify new product
opportunities, as well as to specify the product features, price, and customer
communication. As the initial concept for the TVcenter-HX emerged, the company
briefly exposed it to key users for their feedback. This concept testing enabled TV to
incorporate the suggestions of potential users.
Before proceeding with the assessment, the evaluators studied data and information
related to the TVcenter-HX project. The project manager was asked to hold a briefing
session to introduce both market and technical data, as well as to present a cursory
financial forecast. The key data in the debriefing included:
Preliminary market data: a description of the marketplace including market
existence, probable market size, and market acceptance. The information was
gathered by archival research; key word searches through various trade
magazines, commercial databases, and reports; in-house information and
personnell; and contacts with a few key users.
Preliminary technical data: a technical appraisal including thetechnical
solution, probable architecture , and technical costs, time, and risks. This
information was largely conceptual and was obtained by searching
11
thetechnical literature; utilizing in-house technical expertise; brainstorming
and creative problem-solving sessions; and reviewing competitors’ product
solutions.
Preliminary business data: a rough financial estimation based on very rough
estimates of sales, costs, and investment required and a rough forecast of risk.
Despite the availabity of both technical and market data, the “first cut” homework
was still marked by ambiguity and uncertainty. The reported data might have been
obtained in a specific environment, such as a developed country, and, therefore may not
be valid for other environments, particularly in developing countries like China,
Korea, and the Association of Southeast Asian Nations. Much of this information is
simply not available in developing countries , and when it was, its reliability was
suspect. Further, in an uncertain and dynamic environment , strategic planning becomes
even more important since the decision could seriously impact the financial
performance of the firm. Since the attributes of the new product project may not exactly
satisfy the firms’ ideal, the decision-makers had to deal with the critical issue of
integrating and balancing different criteria. The CEO expressed a desire to pursue a
method that takes into account the uncertainty of each factor yet maintained the nature
of multiplicity to provide an overall picture of the possible success of the TVcenter-HX
development. Since experts can easily differentiate between high, medium, and low, but
find it difficult to judge whether a value, e.g. 0.2, is low, or another value, e.g. 0.3,
is also low, they have found it easier to use linguistic terms to measure ambiguous
events. Since linguistic variables contain ambiguity and multiplicity of meanings and
the information obtained can be expressed as a range in fuzzy set, instead of a single
value in traditional methods, we suggested applying fuzzy logic to this decision
making context.
12
2) Select criteria for decision making: The next step in the screening process is to
decide on the criteria to evaluate the proposed product. A new product Go/NoGo
decision depends not only on the characteristics of the product but also on the
technological competencies and the competitive environment of the company. Since the
situation varies from product to product, there is a high probability that no single set
of factors reflects all situations and requirements even in the same firm. Furthermore,
evaluators with different functional perspectives bring particular needs and desires into
the decision. In order to accurately elicit assessment criteria reflecting the entire set of
attributes of the NPD, the committee proceeded through a series of discussions,
focusing primarily on the nature of the marketplace, competitive circumstances,
technological opportunities, customer requirements, complexity of products/processes,
and the company’s strategy, capabilities and resources.
After the discussion and referring to assessment factors proposed in previous
studies [5], [21], [35]-[38], the team developed a selection architecture and
categorized criteria into four groups: (1) product-marketing competitive advantages: fit
with the company’s core marketing competencies and potential competitive advantage,
(2) product superiority: special features or traits that offer a superior value to users
relative to competitors, (3) technological appropriateness: fit with company’s core
technological competencies so as to bringing about a developing suitability,??? and (4)
product risk: overall level of management uncertainty regarding the project’s
outcomes.
Using the architecture, they further developed/selected sub-criteria for
measurement. Delphi iterative procedures were used to facilitate a consensus on the
selection of different sub-criteria and their relative importance to the firm., Each
primary crierion was expanded into a detailed set of secondary criteri. For example,
competitive marketing advantage was expanded to desired entry timing, offered price
13
level, fit with sales force, distribution channels and logistical strength, and marketing
attractiveness, as shown in Table 2. (Table 2 merely presents what we assess to be the
most prevalent and meaningful factors for this case study).
3) Define linguistic variables and associated membership functions: The ad hoc
usage of linguistic terms and corresponding membership functions is characteristic of
fuzzy logic. It is notable that many popular linguistic terms and corresponding
membership functions have been proposed for linguistic assessment [22], [39]. For the
sake of convenience, instead of eliciting linguistic terms and corresponding
membership functions from the experts, they couldcould be obtained directly from
past data or basic models can be modified to incorporate individual situations and the
requirements of different users. Furthermore, due to limited short-term memory
capacity, it is suggested that the number of linguistic levels not exceed nine.
As the assessment proceeded, the committee members further investigated the new
product attributes, the organization’s capabilities, its marketing ability, its competition,
and the NPD project-related information and data.At first, the managers were unable to
reach a consensus on linguistic variables and membership functions. In order to limit
debate and argument, the linguistic terms and corresponding membership functions
used in previous studies were adopted as and modified to incorporate the specific
requirements of TV. To validate that these linguistic variables and the membership
functions were appropriate and to ease communications within committee, we asked
each of the four evaluators to describe the membership functions when we gave them a
linguistic variable. This continued until their answers reached consensus.
For evaluating the rating effect of the different criteria of the product-marketing
competitive advantages and product superiority, the committee selected the rating scale
R= {Worst [W], Very Poor [VP], Poor [P], Fair [F], Good [G], Very Good [VG], Best
[B]} and its associated membership function as shown in Figure 3. The rating scale R'
14
= {Extremely High [EH], Very High [VH], High [H], Fairly High [FH], Medium [M],
Fairly Low [FL], Low [L]} and its associated membership function as shown in Figure
4, was used for estimating the rating possibility of the different criteria for new-product
ris. The weighting scale W = {Very Low, Low, Fairly Low, Fairly High, High, Very
High} and its associated membership function as shown in Figure 5, for evaluating the
relative importance of the various criteria.
4) Assess the criteria rating and weight using linguistic terms: Once the linguistic
variables and associated membership functions for evaluating the merit ratings and
the importance weights of the selected criteria were defined, the experts used the
linguistic terms to directly assess the rating which characterizes the degree of the
effect/impact of various factors on the success of the new product. Table 3 shows the
results of the assessment under the thirteen criteria given by evaluators E1, E2, E3 and E4,
respectively. Concurrently, the experts evaluated the relative importance of each
criterion by comparison, on the basis of their experience and knowledge. The results are
shown in Table 4.
5) Translate the linguistic ratings and weights into fuzzy numbers: On the basis of
Figure 3 and Figure 4, the linguistic terms of the effect ratings of the thirteen criteria
assessed by each evaluator shown in Table 3 were approximated by fuzzy numbers
parameterized by quadruples, as shown in Table 5. Similarly, on the basis of Figure 5,
the linguistic terms of the importance weighting shown in Table 4 were approximated
by fuzzy numbers and parameterized by quadruples, as shown in Table 6.
6) Aggregate fuzzy numbers to obtain fuzzy merit-importance indexes of selected
criteria and a fuzzy-possible-success-rating: It is important to aggregate the different
experts' opinions in group decision-making. Many methods can be used to aggregate
the experts' assessments, such as mean, median, maximum, minimum, and mixed
operators. Since the median operation is more robust in a small sample, this method
15
was chosen to pool the experts' assessments. The median fuzzy numbers of the effect
ratings shown in Table 5 and the importance weights shown in Table 6 were derived
Fuzzy-possible-success-rating (FPSR) is an information measue which consolidates
fuzzy ratings and fuzzy weightings of all the factors that will influence or impact the
success of the NPD project. It represents the overall merit or attractiveness of an NPD
project. The higher the FPSR of an NPD project is, the stronger the degree of success
for this NPD project. Thus, the membership function of FPSR will be used to determine
an NPD project’s Go/NoGo decision.
Let Rj and Wj, j = 1, 2, …, n, denote the median effect rating and median importance
weighting assigned to factor j, respectively, by the evaluating committee. By
integrating the favorable and unfavorable factors according to the fuzzy
weighted-average definition [41], the fuzzy-possible-success-rating is defined as:
Phpj j
Pi i
Phpj jj
Pi ii WWWRWRFPSR 111
'1 )()( (1)
where p + h = n, and R'j = (1, 1, 1) Rj , j = 1 + p, 2 + P,…, h +p, Rj are the possibility
ratings of the factors of new-product risk. These factors will impact the success of an
NPD project.
Several methods have been devised for calculating the membership function of
fuzzy weighted averages [41]-[44]. In term of the efficiency for calculating the
membership function, the fractional programming approach developed by Kao and Liu
[44] is adopted.
Let wi and ri be positive real numbers (since Wi and Ri are restricted to positive
fuzzy numbers in real world applications) at a specific α-cut of Wi and Ri, respectively.
Following the variable transformation of Charnes and Cooper [45] by letting
ni iwt 11 and vi = twi, the lower and upper bounds of FPSR can be transformed to
the conventional linear program and solved using the following formulation:
16
)(.min1
RvyY iL
n
ii
L
s.t. ,)()( wtvwt iiU
i
L
I = 1, …, n (2a)
n
iiv
1
1
t, vi ≥ 0
)(.max1
RvyY iU
n
ii
U
s.t. ,)()( wtvwt iiU
i
L
I = 1, …, n (2b)
n
iiv
1
1
t, vi ≥ 0
By enumerating different α values, the membership function FPSR can be
constructed. Using the expressions (1), (2a) and (2b) the fuzzy-possible-success-rating
for the TV center-HX development was obtained as:
FPSR = (0.439, 0.666, 0.852)
Furthermore, the fuzzy merit-importance index (FMII), which combines the merit
and importance of each criterion, represents an effect which will contribute to the
success of an NPD project. The lower the FMII of a factor is, the lower the degree of
contribution for this factor. Thus, the FMII score of a factor is used for identifying the
principal adverse factors.
If one uses Figure 5 directly, the fuzzy numbers for approximating the linguistic
values in weighting set W, the importance weightings will neutralize the effect ratings.
Therefore, one cannot identify the actual adverse factors (low rating and high
weighting). Hence, for favorable factors the FMII is defined as:
FMIIi = Ri W'i (3)
17
where W'i = (1,1,1) Wi. , I = 1, 2,..., P
For unfavorable factors the FMII is defined as:
FMIIj = R'j W'j (4)
where W'j = (1,1,1) Wj, j = 1 + p, 2 + P,…, h +p
By using the formulas in Eq (3) and (4), the fuzzy merit-importance index of each
criterion was obtained as listed in Table 7.
7) Translate the fuzzy-possible-success-rating into an appropriate linguistic term:
Once the proposed product’s fuzzy-possible-success-rating has been obtained, one can
further approximate a linguistic label whose meaning is the same as (or closest to) the
meaning of the FPSR from the natural-language expression set of possible success (PS)
for guiding a manager to make a Go/NoGo decision,. Several methods for translating
the membership function back to linguistics have been proposed [46], [47]. There are
basically three techniques: (1) Euclidean distance, (2) successive approximation, and (3)
piecewise decomposition. It is recommended that the Euclidean distance method be
utilized because it is the simplest to implement and the most intuitive form of human
perception of closeness of proximity [48].
The Euclidean method consists of calculating the Euclidean distance from the given
fuzzy number to each of the fuzzy numbers representing the natural-language
expressions set. Assume natural-language expression set PS; then the distance between
the fuzzy number FPSR (known) and each fuzzy number member PSi (unknown) PS
can be calculated as below:
(5)
where p = {x0, x1, …, xm} [0, 1] so that 0 = x0 x1 … xm = 1. Let p = {0, 0.05,
0.1, 0.15, 0.2, 0.25, 0.3, 0.35, 0.4, 0.45, 0.5, 0.55, 0.6, 0.65, 0.7, 0.75, 0.8, 0.85, 0.9,
px
xfPSi
xf FPSRPSFPSRd i
2),(
21
18
0.95, 1}. Then, the distance from the FPSR to each of the members in the set PS can be
calculated, and the closest natural expression with the minimum distance can be
identified.
In this case, the natural-language expression of possible success set, PS = {Very
Low, Low, Fairly Low, Fairly High, High, Very High}, was chosen for labeling, and the
linguistics and corresponding membership functions were shown in Figure 6. Then, by
using formula (5), the Euclidean distance D from the FPSR to each member in set PS
was calculated:
D(FPSR, VL) = 2.1998, D(FPSR, L) =2.1998, D(FPSR, FL) =1.983,
D(FPSR, FH) = 1.3803, D(FPSR, H) = 0.7582, D(FPSR, VH) = 2.0196.
Thus, by matching a linguistic label with the minimum D, the possible success of
the TVcenter-HX development appeared to be High.
In translating the fuzzy-possible-success-rating back into linguistic terms, one can
choose other labels and membership functions, depending on one’s experience and
needs. It is notable that one must choose an even-level???? natural-language expression
of possible-success set. If one chooses an odd-level expression, a “medium” or “fair”
result may be obtained. In such a situation, one cannot give any guidance to the
decision-maker. Hence, it is recommended that one choose an even-level expression.
8) Rank fuzzy merit-importance indices of criterion: As mentioned in the previous
section, a screening evaluation not only determines the NPD Go/NoGo but also, most
importantly, helps managers assess distinctive competencies and identify the principal
adverse factors for proactively implementing appropriate preventive measures. In order
to identify the principal adverse factors for the success of an NPD project, the FMII of
each criterion must be ranked. Many methods have been developed to rank fuzzy
numbers [39], [49]. Here, the ranking of the fuzzy number is based on Chen and
19
Hwang’s left-and-right fuzzy-ranking method, since it not only preserves the ranking
order but also considers the absolute location of each fuzzy number [39].
In this ranking method, the fuzzy maximizing and minimizing sets are, respectively,
defined as:
maxf
,
,x
x x
0 1
0 otherwise (6)
minf
,
,x
x x
1 0 1
0 otherwise (7)
When given a triangular fuzzy number M defined as: f M: R [0, 1] with a
triangular membership function, the right-and-left fuzzy merit-importance index of M
can be obtained, respectively, as:
xfxfMFMII Mx
R maxsup (8)
xfxfMFMII Mx
L minsup (9)
Finally, the fuzzy merit-importance index of M can be obtained by combining the
left and right. This index is defined as:
21 MFMIIMFMIIMFMII LR (10)
By using the ranking method presented in Eq. (6)-(10), the scoring values for the
fuzzy merit-importance indices of the thirteen key success factors were obtained. The
ranking values are shown in Table 7.
Although the possibility of machining-center development was High (according to
the evaluation), there were obstacles within the organization which could have
impacted the success of the project. Using the Pareto principle, the committee decided
20
to focus resources on a few critical factors and set a scale 0.10 as the management
threshold for identifying the critical factors for improvement. Subsequently, as shown
in Table 7, three factors had merit values lower than the threshold, namely: (1) market
competitiveness, (2) marketing attractiveness, and (3) product entry-marketing timing.
These factors represented the most significant contributions for enhancing the success
possibility of the machining-center TVcenter-HX development.
D. Comparison Study
Since the FSLM is an extension of the MCDM approach, in order to ascertain the
efficiency of this method, a comparison study of the and the MCDM approach was
made by the evaluation committee.
When using the MCDM approach for product screening, the ambiguity and
multiplicity within factors are ignored. The evaluators were asked to use a scale to
score the criteria directly orto use linguistic terms to assess the criteria.
Subsequently, the linguistic terms were translated into a crisp scale for computing the
possible-success-rating of the new-product. In the comparison study, we used the
“core” member of the fuzzy number to represent a linguistic value in the MCDM
approach. For example, the triangular fuzzy number (0.5, 0.65, 0.8) was used to
approximate the linguistic variable “Good”, therefore the core member 0.65 was
adopted to represent the linguistic variable “Good” in the MCDM approach. The
contrasting fuzzy numbers for approximating linguistic variables and crisp scales
representing linguistic variables are listed in Table 8.
The results were compared with those derived from the fuzzy logic screening model,
listed in Table 9. As shown in the possible-success-rating scale in Table 9, the results
generated by both approaches seemingly lead to similar conclusions. However, the
21
possible-success rating generated by the FLSM approach is expressed in terms of
ranges of value. This rating can provide an overall picture of the relevant possibility
and ensure that the decision made in the subsequent selection process is not biased.
Further, it allows the managers a high degree of flexibility in decision-making. In the
example in this study, the possible-success rating had a fuzzy value (0.439, 0.666,
0.852). Qualitatively, this suggests that the proposed product is success-high and far
from being a failure. However, a crisp rating of 0.666 generated by MCDM approach
may imply differently or provide less rich information.
E. Go/No-Go Decision
In the TV case study. the analysis showed that the success possibility of the
TVcenter-HX development was high, it had a success rating of 0.439-0.852, far from
being a failing product. After a reconfirming discussion, the committee made a
recommendation that the TVcenter-HX was a worthy selection for development on the
basis of the possible-success-rating of the project. In connection with the weakest
factors within the organization, the committee suggested that an action plan be
conducted to improve adverse factors and to stimulate the possibility of success for the
TVcenter-HX development.
V. CONCLUSIONS AND FUTURE DIRECTIONS
This research has highlighted the importance of product screening in new product
development Because of complexity, incomplete information and ambiguity in the
screening context, a fuzzy logic screening model which applies linguistic
approximation and fuzzy arithmetic has been developed to address new product
22
Go/NoGo decisions. The method incorporates the multiplicity in meaning and
ambiguity of factor measurement while allowing for the consideration of important
interactions among decision levels and criteria. The company and managers involved in
the case study illustrated in this study were generally pleased with the approach. This
study has provided potential value to practitioners by offering a rational structure for
reflecting the imprecise phenomena in many business environments and has taken into
account the uncertainty of each factor to assure a relatively realistic and informative
evaluation, and to researchers by demonstrating another application of fuzzy logic.
Although the case study has demonstrated the usefulness of the model as an
extension to MCDM in new-product screening, it may be very valuable for a company
to use both the NewProd insturment[21] and the fuzzy approach, because each uses
different theoretical approaches and algorithms for new product screening. Further,
believing there are areas for future validation and improvement, we hope to encourage
additional managers to adopt our method. A single case study or a number of case
studies does not necessarily provide a true measure of the relative performance and
success of this model. Further research needs to done bring this model to maturity and
to compare the efficiency of the model in different types of new-product development
selections (such as breakthrough product, new core product, additions to product
families, etc). Another aspect of future research could be to extend this model for use in
a portfolio-selection environment where synergies and overlaps among products
portfolio could be more thoroughly considered.
It is acknowledged that the evaluation levels and members involved in any
particular implementation will be different, depending on the firm involved. The
situations and requirements vary from product to product and from firm to firm. For
example, firms in high tech industries, stressing competitive advantage through
innovation, may have decided on criteria and weighting different from firms in mature
23
industries seeking to compete as low-cost providers of proven technology. In addition,
a model cannot consider all success-enabled factors [20]. We want to emphasize that
the thirteen critical success-enabling attributes are by no means exhaustive; therefore,
new factors may be added/amended, depending on the product, industry and market
characteristics. Future research should examine different models to validate and
compare their efficiency.
Finally, there are some limitations to the fuzzy logic approach. The
membership function of natural-language expression depends on the managerial
perspective of the decision-maker. The decision-maker must be at a strategic level in
the company in order to realize the importance and trends of all aspects, such as
strategy, marketing and technology. Further, competitive situations and requirements
vary from company to company; hence, companies must establish their unique
membership function appropriate to their specific environment and considerations. In
addition, the computation of a fuzzy-weighted average is still complicated and not
easily appreciated by managers. Fortunately, this calculation has been computerized to
effectively reduce the tediousness and time-consuming.
ACKNOWLEDGMENTS:
The authors wish to express appreciation to the editor and three anonymous
reviewers for their valuable suggestions, Dr. Cheryl Rutledge for her editorial
assistance, the company managers for their cooperation and the National Science
Council of Taiwan for its financial support (NSC 90-2218-E-212-014)
24
REFERENCES
[1] G. L. Urban and J. D. Hauser, Design and Marketing of New Products. Englewood
Cliffs, NJ: Prentice Hall, 1993.
[2] M. Ozer, “A survey of new product evaluation models,” Journal of Product
Innovation Management 16, pp. 77-94, 1999.
[3] A Griffin, “PDMA research on new product development practices: Updating
trends and benchmarking best practices,” Journal of Product Innovation
Management 14, pp. 429-458, 1997.
[4] G. R. Cooper and E. J. Kleinschmidit, “An investigation into the new product
process: Steps, deficiencies, and impact,” Journal of Product Innovation
Management 3, pp. 71-85, 1986.
[5] R. J. Calantone, C. A. Di Benedetto and J. B. Schmidt, “Using the analytic
hierarchy process in new product screening,” Journal of Product Innovation
Management 16, pp. 65-76, 1999.
[6] R. G. Cooper, Winning at New Product: Accelerating the Process from Idea to
Launch. Cambridge, Ma: Addison-Wesley, 1993.
[7] R. Balachandra, “Critical signals for making Go/NoGo decisions in new product
development,” Journal of Product Innovation Management 1, pp. 92-100, 1984.
[8] J. B. Schmidt and R. J. Calantone, “Are really new product development projects
harder to shut down?” Journal of Product Innovation Management 15, pp. 429-458,
1997.
[9] H. Ahn and H. Dyckhoff, “A flexible concept for decision support: Evaluation and
selection of suitable product development activities,” European Journal of
Operational Research 100, pp. 266-276, 1997.
[10] G. R. Cooper, S. J. Edgett and E. J. Kleinschmidit, “New product portfolio
management: Practices and performance,” Journal of Product Innovation
Management 16, pp. 333-351, 1999.
[11] L. L. Machacha and P. Bhattacharya, “A fuzzy logic-based approach to project
25
selection,” IEEE Transactions on Engineering Management Vol. 47, No. 1, pp.
65-73, 2000.
[12] A. D. Henriksen and A. J. Traynor, “A practical R&D project-selection scoring
tool,” IEEE Transactions on Engineering Management Vol. 46, No. 2, pp. 158-169,
1999.
[13] R. Kangari and L. S. Riggs, “Construction risk assessment by linguistics,” IEEE
Transactions on Engineering Management Vol. 36, No. 2, pp. 126-131, 1989.
[14] C. A. Di Benedetto, “Defining markets and users for new technologies,” In
Managing New Technology Development. W. E. Isouder and J. D. Sherman (eds.),
New York: McGraw-Hill, 1994.
[15] A. Rangaswamy and G. L. Lilien, “Software tools for new product development,”
Journal of Marketing Research 34, pp. 177-184, 1997.
[16] E. Kahn, Barbara and R. J. Meyer, “Consumer multi-attribute judgment under
attribute weight uncertainty,” Journal of Consumer Research 17, pp. 508-522,
1992.
[17] A. L. Page, “Developing an effective concept testing program for consumer
products,” Journal of Product Innovation Management 9, pp. 267-277, 1992.
[18] M. J. Liberatore, “An extension of the analytic hierarchy process for industrial
R&D project selection and resource allocation,” IEEE Transactions on
Engineering Management EM-34, pp. 12-18, 1987.
[19] A. de Korvin and R. Kleyle, “Fuzzy analytical hierarchical processes,” Journal of
Intelligent & Fuzzy Systems 7, pp. 387-400, 1999.
[20] L. M. Meade and A. Presley, “R&D project selection using the analytic network
process,” IEEE Transactions on Engineering Management Vol. 49, No. 1, pp.
59-66, 2002.
[21] R. G. Cooper, “Selecting winning new product projects: Using the NEWPROD
system,” Journal of Product Innovation Management 2, pp. 34-44, 1985.
[22] W. Karwowski and A. Mital, “Applications of approximate reasoning in risk
analysis,” in Applications of Fuzzy Set Theory in Human Factors. W. Karwowski
26
and A. Mital (eds.), Netherlands, Amsterdam, 1986.
[23] J.Yen and R. Langari, Fuzzy Logic: Intelligence, Control and Information. Upper
Saddle River, NJ: Prentice_hall, 1999.
[24] K. Chen and N. Gorla, “Information system project selection using fuzzy logic,”
IEEE Transactions on Systems, Man, and Cybernetics-Part A: Systems and
Humans Vol. 28, No. 8, pp. 849-855, 1998.
[25] T. Ramsha Prabhu and K. Vizayakumar, “Technology choice using FHDM: A
case of iron-making technology,” IEEE Transactions on Engineering Management
Vol. 48, No. 2, pp. 209-222, 2001.
[26] M. Ben Ghalia and P. P. Wang, “Intelligent system to support judgmental business
forecasting: the case of estimating hotel room demand,” IEEE Transactions on
Fussy Systems Vol. 8, No. 4, pp. 350-397, 2000.
[27] C. Kao and S. T. Liu, “Competitiveness of manufacturing firms: an application of
fuzzy weighted average,” IEEE Transactions on Systems, Man, and
Cybernetics-Part A: Systems and Humans Vol. 29, No. 6, pp. 661-667, 1999.
[28] C. T. Lin, “A knowledge-based method for bid/no-bid decision making in project
management,” Proceedings of PMI Research Conference 2000 Paris, France:
347-355, 2000.
[29] L. H, Chen and T. W. Chiou, “A fuzzy credit-rating approach for commercial
loans: a Taiwan case,” Omega 27(4), 407-419, 1999.
[30] P. C. L. Hui, K. C. C. Chan, K. W. Yeung and F. S. F. Ng, “Fuzzy operator
allocation for balance control of assembly lines in apparel manufacturing,” IEEE
Transactions on Engineering Management Vol. 49, No. 2, pp. 173-180, 2002.
[31] R. A. Roussel, K. N. Saad and J. J. Erickson, Third Generation R&D: Managing
the Link to Corporate Strategy. Boston, MA: Harvard Business Press, 1991.
[32] D. Dubois and H. Prade, Possibility Theory-An Approach to Computerized
Processing of Uncertainty. New York: Plenum, 1988.
[33] R. Muramatsu, T. Ichimura and K. Ishii, “An analysis of needs assessment and
information behavior in product development based on the fusion model,”
27
Technovation 10 (5), pp. 305-317, 1990.
[34] M. Tuominen, P. Piippo, T. Ichimura and Y. Matsumoto, “An analysis of
innovation management systems’ characteristics.” Int. J. Production Economics
60-61, pp.135-143, 1999.
[35] R. G. Cooper, “Predevelopment activities determine new product success,”
Industrial Marketing Management 17, pp. 237-247, 1988.
[36] X. M. Song and M. E. Parry, “What separates Japanese new product winners from
losers?” Journal of Product Innovation Management 13, pp. 422-439, 1996.
[37] C. A. Di Benedetto, “Identifying the key success factors in new product launch,”
Journal of Product Innovation Management 16, pp. 530-544, 1999.
[38] C. M. Crawford and C. A. Di Benedetto, New Product Management, New York
Irwin: McGraw-Hill, 2000.
[39] S. J. Chen and C. L. Hwang, Fuzzy Multiple Attribute Decision Making Methods
and Applications. Berlin Heidelberg: Springer-Verlag, 1992.
[40] C. H. Juang, X. H. Huang and D. J. Elton, “Fuzzy information processing by the
Monte Carlo simulation technique, Journal Civil Engineering Systems 8 (1), pp.
19-25, 1991.
[41] T. S. Liou and M. J. Wang, “Fuzzy weighted average: an improved algorithm,”
Fuzzy Sets and Systems 49, pp. 307-315, 1992.
[42] D. H. Lee and D. Park, “An efficient algorithm for fuzzy weighted average,”
Fuzzy Sets and Systems 87, pp. 39-45, 1997.
[43] H. K. Chen, W. K. Hsu and W. L. Chiang, “A comparison of the vertex method
with the JHE method,” Fuzzy Sets and Systems 95, pp. 201-214, 1998.
[44] C. Kao and S. T. Liu, “Fractional programming approach to fuzzy weighted
average,” Fuzzy Sets and Systems 120, pp. 435-444, 2001.
[45] A. Charnes and W. W. Cooper, “ Programming with linear fractional functionals,”
Nav. Res. Logist. Quart. Vol. 9, pp. 181-186, 1962.
[46] F, Eshragh and E. H. Mandani, “A general approach to linguistic approximation,”
International Journal of Man-Machine Studies 11, 501-419, 1979.
28
[47] K. J. Schmucker, Fuzzy sets, natural language computations and risk analysis.
USA: Computer Science Press, 1985.
[48] H. W. Guesgen and J. Albrecht, “Imprecise reasoning in geographic information
systems,” Fuzzy Sets and Systems 113, pp. 121-131, 2000.
[49] H. Lee-Kwang and J. H. Lee, “A method for ranking fuzzy numbers and its
application to decision-making,” IEEE Transactions on Fuzzy Systems Vol. 7, No.
6, pp. 677-685, 1999.
29
Fig. 1. Evaluation framework of the Fuzzy Logic Screening Model
Selection of
assessment criteria
and evaluation terms
Linguistic evaluation
Translation of
linguistic variables
Aggregation and
inferences of fuzzy
numbers
Criteria and linguistic scales
for evaluation
Linguistic values
Fuzzy numbers
Ranking of
fuzzy numbers
Linguistic
term matching
Fuzzy merit-importance
indices of criterion
Fuzzy
possible-success
ratings
Linguistic
label bank
Management
threshold
Go/No-Go decision and
preventive action planning
Adverse
factors
Go/No-Go
suggestion
Change in
business
environment
Company’s
strategies and
managerial goals
Company’s
competency
and resources
30
Fig. 2. Basic Architecture for New Product Screening
Viable and profitable
new-product ideas:
Right product
features and
characteristics
Right time to
develop
Right amount of
development
investments
Strategies for new-product
screening:
Customer’s values,
expectations and
requirements
Competitive situation
and trend
Company’s goals and
competitive strategies
Technological
opportunities, and
company’s capabilities
and resources
New-product
idea
New-product
Go/NoGo
analysis
Inferior
product idea
Pass
Reject
31
Fig. 3. Fuzzy numbers for approximating linguistic-effect rating values.
(Worst (0, 0, 0.2); Very Poor (0, 0.2, 0 .4); Poor (0.2, 0.35, 0.5); Fair (0.3, 0.5, 0.7);
Good (0.5, 0.65, 0.8); Very Good (0.6, 0.8, 1.0); Best (0.8, 1.0, 1.0)
1.0
F(x)
0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0
x
Worst
Very
Poor
Very
Good Poor Fair Good Best
32
Fig. 4. Fuzzy numbers for approximating linguistic-possibility rating values.
(Low (0, 0, 0.2); Fairly Low (0, 0.2, 0.4); Medium (0.2, 0.35, 0.5);
Fairly High (0.3, 0.5, 0.7); High (0.5, 0.65, 0.8); Very High (0.6, 0.8, 1.0);
Extremely High (0.8, 1.0, 1.0)
1.0
F(x)
0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1.0
x
Low Fairly
Low
Very
High Medium Fairly
High High Extremely
High
33
Fig. 5. Fuzzy numbers for approximating linguistic weighting values.
(Very Low (0, 0, 0.2); Low (0, 0.2, 0.4); Fairly Low (0.2, 0.4, 0.6);
Fairly High (0.4, 0.6, 0.8); High (0.6, 0.8, 1.0); Very High (0.8, 1.0, 1.0);
1.0
F(x)
0 .1 .2 .3 .4 .5 .6
.7 .8 .9 1.0 x
Very
Low
Fairly
High Low
Fairly
Low High
Very
High
34
FPSR
Fig. 6. Linguistic terms for matching fuzzy-possible-success rating value.
(Very Low (0, 0.15, 0.3); Low (0.15, 0.3, 0.45); Fairly Low (0.3, 0.45, 0.6);
Fairly High (0.4, 0.55, 0.7); High (0.55, 0.7, 0.85); Very High (0.7, 0.85, 1.0)
1.0
F(x)
0 .1 .2 .3 .4 .5 .6
.7 .8 .9 1.0 x
Fairly
High Low
Fairly
Low High
Very
High
Very
Low
35
TABLE I
THE PRO AND CON OF MCDM, AHP AND REGRESSION MODELS
Approach Pro Con
Multicriteria
decision
making
techniques
[16], [17]
Easier to implement
and understand for
systematizing the
review of projects
Focus attention on
the most important
issues
Do not take into account the uncertainty
associated with the mapping of one’s
judgment to a number; and
Subjective judgment, selection and
preference of evaluators have a
significant influence on these methods.
Analytic
Hierarchy
Process
[18]-[20]
Reconcile different
managerial
judgment and
perceptions,
Does not account for the uncertainty
associated with the mapping of one’s
judgment to a number,
Subjective judgment, selection and
preference of evaluators have a
significant influence on results.
Used primarily in a selection situation.
May lead to selection of the best in a set
of bad alternatives.
Screening
regression
models [21]
Comprehensive and
useful tool.
Historical database may no longer be
current.
Experience and judgment of one firm
may not be applicable to another firm.
Market success is the only criterion
accounted.
Criteria are all subjective.
Cannot be customized.
36
TABLE II
PRODUCT EVALUATION AND SELECTION CRITERIA
Criteria Description
Competitive
marketing
advantages
Market timing (C11) Matches desired entry timing needed by target
segments
Price superiority
(C12)
Offers value for money to target segments
Marketing
competencies (C13)
Fits in with our salesforce, channels of
distribution and logistical strengths
Marketing
attractiveness (C14)
Permits the company to enter into a growing,
high potential market
Superiority Functional
competency (C21)
Has unique or special functions to meet and
attract target segments
Featured differential
(C22)
Has unique or special features to attract target
segments
Technological
suitability
Design quality (C31) Is designed for the quality needed by target
segments
Material
specialization (C32)
Uses materials of high quality and low rejection
Manufacturing
compatibility (C33)
Can be produced by our best manufacturing
technology and flexibility
Supply benefit (C34) Allows the company to use very best suppliers
Risk Market
competitiveness (C41)
Many competitive products in the market
Technological
uncertainty (C42)
Uses new technological skills that cannot be
addressed by research
Monetary risk (C43) Total dollar risk profile of product
37
TABLE III
EFFECT AND POSSIBILITY RATINGS OF CRITERIA ASSIGNED BY EXPERTS
USING LINGUISTIC TERMS
Experts Criteria
C11 C12 C13 C14 C21 C22 C31 C32 C33 C34 C41 C42 C43
E1 G F F VG B VG VG G B F H H M
E2 B G P VG B VG VG VG VG G VH VH H
E3 B P P B VG B VG G VG F H H FH
E4 VG F F B B VG B VG G G VH H M
TABLE IV
IMPORTANCE WEIGHTINGS OF CRITERIA ASSESSED BY EXPERTS USING
LINGUISTIC TERMS
Expert Criteria
C11 C12 C13 C14 C21 C22 C31 C32 C33 C34 C41 C42 C43
E1 VH FL VH H VH FH H H H FH VH H FH
E2 H H VH VH H FL H FH FH H H H H
E3 VH H H VH VH FH VH FH FL FH VH VH FH
E4 VH FH H VH H FH VH FL FH FH VH H FL
41
TABLE V
EFFECT RATINGS OF CRITERIA APPROXIMATED BY FUZZY NUMBERS
Criteria
Experts
E1 E2 E3 E4 Median
C11 (0.5, 0.65, 0.8) (0.8, 1.0, 1.0) (0.8, 1.0, 1.0) (0.6, 0.8, 1.0) (0.7, 0.9, 1.0)
C12 (0.3, 0.5, 0.7) (0.5, 0.65, 0.8) (0.2, 0.35, 0.5) (0.3, 0.5, 0.7) (0.3, 0.5, 0.7)
C13 (0.3, 0.5, 0.7) (0.2, 0.35, 0.5) (0.2, 0.35, 0.5) (0.3, 0.5, 0.7) (0.25, 0.43, 0.6)
C14 (0.6, 0.8, 1.0) (0.6, 0.8, 1.0) (0.8, 1.0, 1.0) (0.8, 1.0, 1.0) (0.7, 0.9, 1.0)
C21 (0.8, 1.0, 1.0) (0.8, 1.0, 1.0) (0.6, 0.8, 1.0) (0.8, 1.0, 1.0) (0.8, 1.0, 1.0)
C22 (0.6, 0.8, 1.0) (0.6, 0.8, 1.0) (0.8, 1.0, 1.0) (0.6, 0.8, 1.0) (0.6, 0.8, 1.0)
C31 (0.6, 0.8, 1.0) (0.6, 0.8, 1.0) (0.6, 0.8, 1.0) (0.8, 1.0, 1.0) (0.6, 0.8, 1.0)
C32 (0.5, 0.65, 0.8) (0.6, 0.8, 1.0) (0.5, 0.65, 0.8) (0.6, 0.8, 1.0) (0.55, 0.73, 0.9)
C33 (0.8, 1.0, 1.0) (0.6, 0.8, 1.0) (0.6, 0.8, 1.0) (0.5, 0.65, 0.8) (0.6, 0.8, 1.0)
C34 (0.3, 0.5, 0.7) (0.5, 0.65, 0.8) (0.3, 0.5, 0.7) (0.5, 0.65, 0.8) (0.4, 0.58, 0.75)
C41 (0.5, 0.65, 0.8) (0.6, 0.8, 1.0) (0.5, 0.65, 0.8) (0.6, 0.8, 1.0) (0.55, 0.73, 0.9)
C42 (0.5, 0.65, 0.8)) (0.6, 0.8, 1.0) (0.5, 0.65, 0.8) (0.5, 0.65, 0.8) (0.5, 0.65, 0.8)
C43 (0.2, 0.35, 0.5) (0.5, 0.65, 0.8) (0.3, 0.5, 0.7) (0.2, 0.35, 0.5) (0.25, 0.43, 0.6)
42
TABLE VI
IMPORTANCE WEIGHTINGS OF CRITERIA APPROXIMATED BY FUZZY NUMBERS
Criteria
Experts
E1 E2 E3 E4 Median
C11 (0.8, 1.0, 1.0) (0.6, 0.8, 1.0) (0.8, 1.0, 1.0) (0.8, 1.0, 1.0) (0.8, 1.0, 1.0)
C12 (0.2, 0.4, 0.6) (0.6, 0.8, 1.0) (0.6, 0.8, 1.0) (0.4, 0.6, 0.8) (0.5, 0.7, 0.9)
C13 (0.8, 1.0, 1.0) (0.8, 1.0, 1.0) (0.6, 0.8, 1.0) (0.6, 0.8, 1.0) (0.7, 0.9, 1.0)
C14 (0.6, 0.8, 1.0) (0.8, 1.0, 1.0) (0.8, 1.0, 1.0) (0.8, 1.0, 1.0) (0.8, 1.0, 1.0)
C21 (0.8, 1.0, 1.0) (0.6, 0.8, 1.0) (0.8, 1.0, 1.0) (0.6, 0.8, 1.0) (0.7, 0.9, 1.0)
C22 (0.4, 0.6, 0.8) (0.2, 0.4, 0.6) (0.4, 0.6, 0.8) (0.4, 0.6, 0.8) (0.4, 0.6, 0.8)
C31 (0.6, 0.8, 1.0) (0.6, 0.8, 1.0) (0.8, 1.0, 1.0) (0.8, 1.0, 1.0) (0.7, 0.9, 1.0)
C32 (0.6, 0.8, 1.0) (0.4, 0.6, 0.8) (0.4, 0.6, 0.8) (0.2, 0.4, 0.6) (0.4, 0.6, 0.8)
C33 (0.6, 0.8, 1.0) (0.4, 0.6, 0.8) (0.2, 0.4, 0.6) (0.4, 0.6, 0.8) (0.4, 0.6, 0.8)
C34 (0.4, 0.6, 0.8) (0.6, 0.8, 1.0) (0.4, 0.6, 0.8) (0.4, 0.6, 0.8) (0.4, 0.6, 0.8)
C41 (0.8, 1.0, 1.0) (0.6, 0.8, 1.0) (0.8, 1.0, 1.0) (0.8, 1.0, 1.0) (0.8, 1.0, 1.0)
C42 (0.6, 0.8, 1.0) (0.6, 0.8, 1.0) (0.8, 1.0, 1.0) (0.6, 0.8, 1.0) (0.6, 0.8, 1.0)
C43 (0.4, 0.6, 0.8) (0.6, 0.8, 1.0) (0.4, 0.6, 0.8) (0.2, 0.4, 0.6) (0.4, 0.6, 0.8)
43
TABLE VII
FUZZY MERIT-IMPORTANCE INDICES OF THIRTEEN CRITERIA
Criterion Rating Weighting Fuzzy merit-importance
index
Ranking
score
C11 (0.7, 0.9, 1.0) (0.0, 0.0, 0.2) (0.0, 0.0, 0.2) 0.083
C12 (0.3, 0.5, 0.7) (0.1, 0.3, 0.5) (0.03, 0.15, 0.35) 0.213
C13 (0.25, 0.43, 0.6) (0.0, 0.1, 0.3) (0.0, 0.043, 0.18) 0.100
C14 (0.7, 0.9, 1.0) (0.0, 0.0, 0.2) (0.0, 0.0, 0.2) 0.083
C21 (0.8, 1.0, 1.0) (0.0, 0.1, 0.3) (0.0, 0.1, 0.3) 0.171
C22 (0.6, 0.8, 1.0) (0.2, 0.4, 0.6) (0.12, 0.32, 0.6) 0.368
C31 (0.6, 0.8, 1.0) (0.0, 0.1, 0.3) (0.0, 0.08, 0.3) 0.160
C32 (0.55, 0.73, 0.9) (0.2, 0.4, 0.6) (0.11, 0.29, 0.54) 0.339
C33 (0.6, 0.8, 1.0) (0.2, 0.4, 0.6) (0.12, 0.32, 0.6) 0.368
C34 (0.4, 0.58, 0.75) (0.2, 0.4, 0.6) (0.08, 0.23, 0.45) 0.284
C41 (0.1, 0.27, 0.45) (0.0, 0.0, 0.2) (0.0, 0.0, 0.09) 0.041
C42 (0.2, 0.35, 0.5) (0.0, 0.2, 0.4) (0.0, 0.07, 0.2) 0.121
C43 (0.4, 0.57, 0.75) (0.2, 0.4, 0.6) (0.08, 0.23, 0.45) 0.284
45
TABLE VIII.
FUZZY NUMBERS FOR APPROXIMATING LINGUISTIC VARIABLES VS. CRISP SCALES REPRESENTING LINGUISTIC VARIABLES
Effect
rating
Linguistic variables Worst Very poor Poor Fairly Good Very Good Best
Fuzzy number (0., 0, 0.2) (0., 0.2, 0.4) (0.2, 0.35, 0.5) (0.3, 0.5, 0.7) (0.5, 0.65, 0.8) (0.6, 0.8, 1.0) (0.8, 1.0, 1.0)
Crisp scale 0. 0.2 0.35 0.5 0.65 0.8 1.0
Possibility
rating
Linguistic variables Low Fairly Low Medium Fairly High High Very High Extremely
High
Fuzzy number (0., 0, 0.2) (0., 0.2, 0.4) (0.2, 0.35, 0.5) (0.3, 0.5, 0.7) (0.5, 0.65, 0.8) (0.6, 0.8, 1.0) (0.8, 1.0, 1.0)
Crisp scale 0. 0.2 0.35 0.5 0.65 0.8 1.0
Importance
weighting
Linguistic variables Very Low Low Fairly Low Fairly High High Very High
Fuzzy number (0., 0, 0.2) (0., 0.2, 0.4) (0.2, 0.4, 0.6) (0.4, 0.6, 0.8) (0.6, 0.8, 1.0) (0.8, 1.0, 1.0)
Crisp scale 0. 0.2 0.4 0.6 0.8 1.0
45
TABLE IX Do you need this?
COMPARISON THE RESULTS OF FLSM AND MCDM APPROACH
Approach Possible-success rating Range Linguistic translation
FLSM (0.439, 0.666, 0.852) 0.413 High
MCDM 0.666