Pengembangan Desain Produk Aromaterapi Khas Bali Berbasis Kansei Engineering Dan Tipe Kepribadian Konsumen

BALINESE AROMATHERAPY PRODUCT DEVELOPMENT
BASED ON KANSEI ENGINEERING AND CUSTOMER
PERSONALITY TYPE

IDA BAGUS DHARMA YOGA SANTOSA

GRADUATE SCHOOL
BOGOR AGRICULTURAL UNIVERSITY
BOGOR
2015

DECLARATION OF ORIGINALITY
AND COPYRIGHT TRANSFER*
Hereby, I declare that the thesis entitled Balinese Aromatherapy Product
Development Based on Kansei Engineering and Customer Personality Type is my
own work under supervision of Dr Eng Taufik Djatna, STP, MSi and Dr Ir Luh
Putu Wrasiati, MP. It has never previously been published in any university. All
of incorporated originated references from other published as well as unpublished
papers are stated clearly in the text as well as in the references.
Hereby, I delegate that the copyright to this paper is transferred to Bogor
Agricultural University.

Bogor, July 2015
Ida Bagus Dharma Yoga Santosa
Student ID F351130251

RINGKASAN
IDA BAGUS DHARMA YOGA SANTOSA. Pengembangan Desain Produk
Aromaterapi Khas Bali Berbasis Kansei Engineering dan Tipe Kepribadian
Konsumen. Dibimbing oleh TAUFIK DJATNA dan LUH PUTU WRASIATI.
Produk aromaterapi khas Bali memiliki suatu keunikan lokal yang membuat
produk ini unik dan terkenal di pasaran. Pada penelitian ini, pendekatan affective
engineering berbasis Kansei Engineering diaplikasikan untuk mengembangkan
produk aromaterapi khas Bali. Tipe kepribadian konsumen berdasarkan teori
Hippocrates dan Galen digunakan pada proses pengembangan produk ini.
berdasarkan motivasi terebut, maka dirumuskan tiga tujuan untuk penelitian ini,
antara lain: (1) menghasilkan konsep desain baru untuk mengembangkan produk
aromaterapi khas Bali, (2) mengidentifikasi elemen-elemen desain produk yang
relevan, dan (3) menghasilkan suatu model kuantifikasi dari desain produk
aromaterapi khas Bali.
Terdapat 12 kata Kansei yang telah dikumpulkan dengan mewawancarai
para pakar, selanjutnya kata-kata tersebut dievaluasi oleh 30 orang responden

menggunakan skala Likert. Metode Principal Component Analysis (PCA)
digunakan untuk menganalisis hasil evaluasi tersebut sehingga dapat
menghasilkan suatu konsep desain baru. Di sisi lain, dari sampel produk yang
telah dikumpulkan, telah teridentifikasi 2 elemen desain untuk komposisi minyak
aromaterapi dan 12 elemen desain kemasannya. Berdasarkan hasil tersebut,
metode Relief digunakan untuk mengidentifikasi relevansi dari elemen desain
tersebut berdasarkan konsep desain yang telah diperoleh. Selanjutnya,
Quantification Theory Type 1 (QTT1) digunakan untuk menganalisis hubungan
antara konsep desain baru dengan elemen desain yang relevan dan Fuzzy
Quantification Type 1 (FQTT1) digunakan untuk menganalisis hubungan tersebut
dengan mempertimbangkan kesan keunikan Bali pada produk tersebut sehingga
dapat menghasilkan suatu model kuantifikasi dari desain produk aromaterapi khas
Bali
Hasil dari metode PCA menunjukkan bahwa terdapat satu komponen utama
yang dapat mewakili keseluruhan komponen yang diberi nama “Bali Tradisional”
(BT). Dengan menggunakan metode Relief, kedua elemen desain komposisi
produk dan 10 elemen desain kemasan produk telah teridentifikasi sebagai elemen
desain yang relevan dengan konsep BT. Berdasarkan hasil dari QTT1 dan FQTT1,
suatu model kuantitatif linear telah dibangun untuk masing-masing tipe
kepribadian. Berdasarkan model ini, suatu informasi pendukung desain telah

diperoleh untuk membantu desainer produk untuk mengembangkan produk
aromaterapi khas Bali. Secara umum, hasil menunjukkan bahwa spesifikasi desain
yang diperoleh bersifat spesifik untuk masing-masing kepribadian dan sebagian
besar dari desain tersebut bersifat produktif sebagai suatu rule.
Kata kunci: produk aromaterapi, Kansei Engineering, tipe kepribadian konsumen,
Quantification Theory Type 1, Fuzzy Quantification Theory Type 1

SUMMARY
IDA BAGUS DHARMA YOGA SANTOSA. Balinese Aromatherapy Product
Design Development Based on Kansei Engineering and Customer Personality
Type. Supervised by TAUFIK DJATNA and LUH PUTU WRASIATI.
Balinese aromatherapy products have such local indigenous that make this
product unique and famous in the market. In this research, affective engineering
approach based on Kansei Engineering is applied to develop Balinese
aromatherapy product. Customer personality type based on Hippocrates and Galen
theory were involved in the development process. According to its main
motivation, three research objectives were formulated (1) to generate the new
design concept of Balinese aromatherapy product, (2) to identify the relevant
product design element, and (3) to generate the quantification model of
aromatherapy product design.

In this research, twelve Kansei words were collected by interviewing the
experts and were evaluated by 30 respondents using Likert scale. Principal
Component Analysis (PCA) were then used to analyze the result and to generate a
new Balinese design concept. On the other hand, from sample products collected,
two design element of mixture composition and twelve design elements of
product container were identified. From the result, Relief method selected and
identified its relevancy to the new design concept. A new design concept,
important design elements, and product Balinese sense were then synthesized
using Fuzzy Quantification Theory Type 1 (FQTT-1) to generate the
quantification model of aromatherapy product design.
The result of PCA method showed that one principal component was
generated to represent a new design concept named “Bali Traditional” (BT). By
using Relief method, all design element of mixture composition and ten design
elements of product container were selected as a relevant design element
correspond to BT concept. According to the result of QTT1 and FQTT1, a linear
quantitative model has been built for each personality type. From these models,
design support information were generated to assist product designers on
developing the new Balinese product in both mixture composition and product
container. In general, the result showed that all design specification of Balinese
aromatherapy product were specific for each personality and most of them were

productive as the new design rules.
Keywords: aromatherapy product, Kansei Engineering, customer personality
type, Quantification Theory Type 1, Fuzzy Quantification Theory Type 1.

© Copyright 2015 by IPB
All Rights Reserved
No Part or all of this thesis may be excerpted without or mentioning the sources.
Excerption only for research and education use, writing for scientific papers,
reporting, critical writing or reviewing of a problem. Excerption doesn’t inflict a
financial loss in the paper interest of IPB.
No part or all of this thesis may be transmitted and reproduced in any forms
without a written permission from IPB.

i

BALINESE AROMATHERAPY PRODUCT DEVELOPMENT
BASED ON KANSEI ENGINEERING AND CUSTOMER
PERSONALITY TYPE

IDA BAGUS DHARMA YOGA SANTOSA


Thesis
as partial fulfillment of the requirements
for the degree of Master of Science
in the Agroindustrial Technology Study Program

GRADUATE SCHOOL
BOGOR AGRICULTURAL UNIVERSITY
BOGOR
2015

ii

Non-committee examiner:

Dr Dwi Setyaningsih, STP, MSi

iii

Thesis Title : Balinese Aromatherapy Product Development Based on Kansei

Engineering and Customer Personality Type
Name
: Ida Bagus Dharma Yoga Santosa
Student ID : F351130251

Approved by
Supervisor

Dr Eng Taufik Djatna, STP, MSi
Chair

Dr Ir Luh Putu Wrasiati, MP
Member

Acknowledge by

Head of Agroindustrial
Technology Study Program

Dean of Graduate School


Prof Dr Ir Machfud, MS

Dr Ir Dahrul Syah, MScAgr

Examination date:
July 9th 2015

Passed date:

iv

v

PREFACE
I would like to thank “Tuhan Yang Maha Esa” for all His gifts so that this
research is successfully completed. The theme chosen in the research which
conducted starting June 2014 up to February 2015 is Kansei Engineering, with
entitled Balinese Aromatherapy Product Development Based on Kansei
Engineering and Customer Personality Type.

I would like to express my sincere gratitude to Dr Eng Taufik Djatna as
Chairman of Advisory Committee for his support and encouragement during my
study in Bogor Agricultural University, to Dr Ir Luh Putu Wrasiati as Member of
Advisory Committee for her advice and supervision during the thesis work, and
also to Dr Ir Titi Candra Sunarti, M.Si as study program representation and Dr
Dwi Setyaningsih, STP, MSi as external examiner for their advice during thesis
examination. Additionally, I would like to thank Direktorat Pendidikan Tinggi
(Dikti) as a sponsorship of my magister study. Then, I would like to say many
thanks to I Wayan Sukhana, Ida Ayu Putu Surya Esti Penida, and Kadek Eka
Citrawati as the experts of this research for their advice, support, and
encouragement in this research.
I would like to say many thanks to my family, my father IB Surya Manuaba,
my mother Annie M Goller, my sister IA Ratih Stefani, and my closest friend
NLPE Kartika Sari for their true and endless love for never failing patience and
encouragement. I would like to thank all lecturers and staff of Agroindustrial
Technology Department, all of my colleagues, especially my best friend in Agroindustrial Technology (Elfira Febriani, Hety H Hidayat, Novi Purnamasari, Aditia
Ginantaka, Nina Hairiyah, Nova Alemina Sitepu, and Elfa S Thamrin), colleagues
in Computer Laboratory of Agro-industrial Technology Department (Iwan Aang,
Zaki Hadi, Riva Aktivia, Rahmawati, Azri Firwan, Rohmah, Fajar
Munichputranto, Husnul Khotimah, Puspa, Aisyah, Yudhis, Ihsan, Imam, Denny),

all of colleagues in Agro-industrial Technology 2013 and 2014, and my
colleagues in KMHD (Joni A Wiranata, Manu Mahatmayana, Perdana Kumara,
Giri Putera, Didi Harizena, Pratyaksa Yudha, AA Keswari Krisnandika, Nengah
Putri Adnyani, Siluh Putu Dia, etc), for their support. It has been a pleasure to
work and sharing with you.
Hopefully this thesis is useful.

Bogor, July 2015
Ida Bagus Dharma Yoga Santosa

vi

TABLE OF CONTENT
TABLE LIST

vii

FIGURE LIST

vii


APPENDIX LIST

viii

1 INTRODUCTION
Background
Problems Statement
Objectives
Benefits
Boundaries

1
1
2
3
3
3

2 RELATED WORK

3

3 METHOD
Research Period and Location
Research Framework
Design Concept of Balinese Aromatherapy Product
Balinese Aromatherapy Product Design Element
Quantification Model of Balinese Aromatherapy Product
Data Analysis
Principal Component Analysis
Relief
Quantification Theory Type I
Fuzzy Quantification Theory Type 1
Fisher Exact Test

4
4
4
6
6
6
7
7
7
8
9
11

4 RESULT AND DISSCUSSION
Design Concept of Balinese Aromatherapy Product
Product domain determination
Design concept generation
Balinese Aromatherapy Product Design Element
Design element of mixture composition
Design element of product container
Quantification Model of Balinese Aromatherapy Product
Personality type of respondent
Product Balinese sense fuzzification
Quantification model of mixture composition
Quantification model of product container
Rule assessment
Advantage and Disadvantage

11
11
11
12
14
14
15
17
17
18
18
21
26
27

5 CONCLUSION AND RECOMMENDATION
Conclusion
Recommendation

27
27
27

REFERENCES

28

vii

TABLE LIST
1
2
3
4
5
6
7
8

Identification of product mixture composition
The result of design element identification
Formulation of aromatherapy product mixture composition based on
customer personality
Design support information for the aromatherapy product design based
on customer personality
Formulation of aromatherapy product container based on customer
personality
The result of design support information of BT concept
The result of Fisher exact test of product mixture composition
The result of Fisher exact test of product container

14
16
19
20
23
25
26
26

FIGURE LIST
1
2
3
4
5
6
7
8

Research method flow chart
The result of eigenvalue
Variables factor map (PCA)
The Relief result of product mixture composition
The Relief result of product container
The proportion of personality type of respondent
The PCC result of product mixture composition
Mixture composition category score: (a) Sanguine design, (b) Choleric
design, (c) Melancholic design, and (d) Phlegmatic design
9 Membership functions of the linguistic term for each design element: (a)
top shape, (b) body shape, (c) bottle height and width ratio, (d) Copped
ornament and bottle height ratio, (e) bottle material, (f) cap color, (g)
ornament item used, (h) root in the bottle, (i) bottle transparency, (j)
label position.
10 Category score: (a) Sanguine design, (b) Choleric design, (c)
Melancholic design, and (d) Phlegmatic design.
11 The PCC result: (a) Sanguine design, (b) Choleric design, (c)
Melancholic design, (d) Phlegmatic design

5
13
13
15
17
18
19
20

22
24
25

viii

APPENDIX LIST
1 Product domain and Kansei words identification questionnaire
2 Product sample collection
3 Kansei words evaluation questionnaire
4 Personality type identification and new design concept evaluation
questionnaire
5 Product Balinese sense evaluation questionnaire
6 The result of product domain identification
7 The result of KW evaluation
8 The result of PCA
9 The result of product mixture composition classification and
evaluation (hypothetical data)
10 The result of product container classification and evaluation
11 The result of Relief
12 The result of customer personality type identification
13 The result of product mixture composition evaluation (hypothetical
data)
14 The result of product Balinese sense evaluation
15 The result of product container design fuzzification
16 The result of product container evaluation for each personality

30
33
38
44
53
55
60
62
63
64
66
67
68
69
70
72

1

1 INTRODUCTION
Background
Balinese aromatherapy products, such as aromatherapy scrub, incense,
massage oil, soap, and many more have such uniqueness which has a big
contribution to make the product become famous in the market (Archam et al.
2013). Moreover these products were coped high value of business since Bali has
chosen as “The Best Spa Destination of The World in 2009” on SENSE Magazine
version (Dispar 2010). In addition, from 2009 to 2013, about 2 803 181 visitors
came to Bali with 9.5% increased for each year (BPS 2014). Considering this
condition, in order to winning the competition in this sector, the need of the
development of aromatherapy product has always been a challenge as well as the
changing of customer demands and expectations.
Currently, Balinese aromatherapy product still applied conventional method
in its product development which beginning from product designer who proposed
the product design, decision maker who decided whether the design accepted or
not, until manufacturing unit to produce the product based on its design. This
conventional method has several disadvantages especially when dealing with the
multidesign product such as Balinese aromatherapy massage oils. These
limitations occurred in such condition (Ko 2010) as right idea rarely comes at the
right moment, the result depends strongly on individual talent and experiences,
and the solution will be limited by one’s special training and experience.
In order to cope these limitations, affective engineering approach based on
Kansei Engineering (KE) was proposed in this research as the improvement of
current conventional method. KE is a product development method that translates
customer impression such as like or dislike, good or bad, and so on (Peter and
Oalson 2005) into product design specification (Nagamachi 2011). By using KE
as a product development method is expected to stimulate the idea generation of
product designer so the result is not depends on individual talent and experiences
anymore. Eventually, the design solution will not limited by one’s special training
and experience anymore because there are customer impression that involved in
the design.
On the previous research, KE has been applied on the development of the
many design products, such as wine glass design (Petiot and Yannou 2004),
rocker switch design (Scutte and Eklund 2005), fragrance packaging design (Wei
et al. 2011), automobile head-up display (Smith and Fu 2011), shampoo
packaging design (Archam et al. 2013), and aromatherapy mixture composition
(Djatna and Santosa 2014). In this research, KE was applied on Balinese
aromatherapy product development by identifying customer needs implicitly and
relates them to product design specification (Lokman 2010).
As mentioned previously, Balinese aromatherapy products have such
uniqueness that make these products become famous in the market. Wijaya (2011)
said that the famous product refers to the product which could fulfill the customer
requirement. However, it is quite hard to fulfill all customer requirement on
aromatherapy product development because these product have very subjective
preferences (Djatna and Santosa 2014). Thus, corresponds to Djatna and Santosa

2

(2014) in order to deal with this condition, customer personality type was
proposed to involve in the product development method. This solution is also
supported by Jordan (2000) who said that each customer has particular criteria on
which they want to buy related to their personality, for instance generally,
extrovert customers prefer gold color than black or grey.
The research challenges in aromatherapy product design are consisted of two
main parts which about how to construct aromatherapy mixture composition and
its container. Refer to Djatna and Santosa (2014), it is proposed Quantification
Theory Type 1 (QTT1) method to formulate the quantification model of
aromatherapy mixture composition. In this research, the quantification model of
aromatherapy mixture composition was generated as a result of analysis of the
relationship between customer impression related to design concept given and
product design elements related to aromatherapy product mixture composition.
Another challenge is about how to develop Balinese aromatherapy product
container while preventing its uniqueness aspect. With considering Balinese sense
of the product as the uniqueness aspect, Fuzzy Quantification Theory Type 1
(FQTT1) is proposed to quantify the design of Balinese aromatherapy product
container. By using FQTT1, quantification model were generated from the result
of relationship between customer impression related to design concept given and
product container design elements with considering Balinese sense of the product.
By using these model generated from both QTT1 and FQTT1 analysis, the
recommendation about combination of the design elements on both mixture
composition and container were obtained directly for each customer personality
type. This recommendation is expected to be the usefull information that able to
assist product designer to develop the Balinese aromatherapy product in order to
fulfill customer subjective preferences.

Problems Statement
It is always been a big challenge for product designer to develop the product
continuously as well as the changing of customer requirement and preference. In
order to assist product designer to develop the product, KE was chosen as a
product development method which involving customer impression on its process.
Moreover customer personality type were involved to pursue the pleasurable
design which reflecting their own personality. Thus, there were several problems
related to Balinese aromatherapy product development based on KE and customer
personality type, as follows:
1. How to generate the design concept of Balinese aromatherapy product based
on its customer impression?
2. How to identify the product design elements which are relevant for the new
design concept generation?
3. How to generate a quantification model of Balinese aromatherapy product in
both mixture composition and its container based on Kansei Engineering
and customer personality type?

3

Objectives
Based on described proposed problem above, there are three main objectives
formulated in this research as follows:
1.
To generate the design concept of Balinese aromatherapy product
2.
To identify the relevant product design elements
3.
To generate the quantification model of Balinese aromatherapy product
design

Benefits
The research is expected to give three benefits which consist of to provide
the information related to the new design concept generation, product design
element which relevant to its design concept, and quantification model of Balinese
aromatherapy product design. Moreover, the result of this research is expected to
give the recommendation about combination of the product design elements for
each customer personality type. This would be useful information to assist product
designer to develop the Balinese aromatherapy product in order to fulfill customer
subjective preferences.

Boundaries
In order to concentrate for solution, there were several boundaries in this
research as follow:
1.
This research was conducted in Bali, so does the sample collected and
respondent involved
2.
Balinese aromatherapy massage oil was chosen as the object of this research
3.
The quantification model of aromatherapy mixture composition were
generated from QTT-1 analysis by using hypothetical data
4.
The quantification model of aromatherapy product container were generated
from FQTT-1 analysis by using survey data

2 RELATED WORK
For years, KE has been applied on the development of the many product
design, such as wine glass design (Petiot and Yannou 2004), rocker switch design
(Scutte and Eklund 2005), fragrance packaging design (Wei et al. 2011),
automobile head-up display (Smith and Fu 2011), and mixture composition of
aromatherapy product (Djatna and Santosa 2014). These four research have been
used Quantification Theory Type 1 (QTT-1) to synthesize the Kansei design
concept and product design element.
Currently, QTT-1 method is usually used in Kansei Engineering research
(Nagamachi 2011) due to its simplicity and its effectiveness related to generate
the linear model of design which represents the optimum combination of product

4

design element and the level of design element affection refers to customer
subjective preference. In this research QTT-1 method were used to generate the
quantification model of aromatherapy product mixture composition for each
personality type.
Additionally, Fuzzy Quantification Theory Type 1 (FQTT-1) were used to
synthesize three dimensions of Kansei which consisted of Kansei design concept,
design element, and Balinese sense of the product. Additionally, design element
fuzzy group were deployed to generate the membership degree of each variance of
design element which is ranged between zero to one. Involving the membership
degree of both design element and product Balinese sense is expected to obtain
the better result of the new design model generation for each personality type.
Based on the model generated from both QTT-1 and FQTT-1 method, the
new design combination were generated which is represented by linear model
approach. From the result, a design support information would be proposed to
give the recommendation about optimum design element combination of the
Balinese design directly based on customer personality type. This would be useful
information for product designer to develop the product especially for Bali
aromatherapy product for each customer personality type in order to minimize the
risk of the failure of the product design in the market.

3 METHOD
Research Period and Location
This research was conducted from June 2014 to February 2015. The sample
products were collected in both modern and traditional market in Bali from July to
August 2014. The survey activities were conducted at Sensory Laboratory of
Agro-industrial Technology, Udayana University, Jimbaran. Thus, data analysis
were conducted at Computer Laboratory of Agro-industrial Technology, Bogor
Agricultural University, Bogor.

Research Framework
The research framework was designed in parallel form to solve the problem
which beginning from problem definition, research objectives, methodology,
result discussion, and conclusion. The methodology was designed to accomplish
the research objectives. In this research, R language version 3.1.3 (Ihaka and
Gentleman 2015) was used as a tool of data processing. The framework detail and
process of the research method is represented in Figure 1.
In this research, generally, based on expert survey and discussion,
hypothetical data were used to generate the quantification model of aromatherapy
product mixture composition. Meanwhile in order to generate the quantification
model of aromatherapy product container, four questionnaires were used to
acquire the data related to product development based on KE and customer
personality type.

5

Start
1st Objective

3rd Objective

Determining product domain
Collecting Kansei word

Identifying customer’s
personality type

2nd Objective

Collecting product sample
Classifying customer based
on their personality

Evaluating Kansei word
collected

Identifying product’s design
element

Generating new product
design concept

Evaluating expert’s Balinese
sense of product samples

Evaluating product samples
Generating fuzzy
membership function of
Balinese sense

Generating the relevant
product design element

New product
design concept

The relevant
design element

Balinese sense
fuzzy group

Mixture composition
Design element
Product container
Design element

Generating fuzzy membership
function of design element

Analyzing by using QTT1
method based on customer
personality types

Fuzzy design
element

Analyzing by using FQTT1
method based on customer
personality types

Partial
correlation
coefficient

Category scores

Generating the highest
score of each design
category

Design support
information
Rule assessment

Productive rule

End

Figure 1 Research method flow chart

Generating the highest
value of PCC

6

Three experts and thirty customers of aromatherapy product were involved
in this research. The experts involved are the owner of Balinese aromatherapy
product and have over ten years of experience in aromatherapy product business.
The respondents were selected as a result of purposive sampling with several
criteria as follows: (1) they have known the product, (2) they have seen the
product, (3) they have used the product, and (4) they will buy the product. Thus,
the evaluation step was conducted by using Likert scale with seven scale.
Design Concept of Balinese Aromatherapy Product
The research method began with the identification of product domain as the
research object. Afterward, the Kansei words (KW) were collected by
interviewing three experts. In this phase, a questionnaire (see Appendix 1) was
given to the experts which aimed to determine the product domain and to collect
the Kansei words. From the result of product domain determination, total amount
25 various samples of Balinese aromatherapy product have been collected from
both traditional and modern market in Bali (see Appendix 2). With involving 30
respondents, the product samples collected and the collection of Kansei words
was then evaluated by using second questionnaire (see Appendix 3). The
evaluation result was then analyzed using Principal Component Analysis to
generate a new Balinese aromatherapy design concept as well as to achieve the
first objective of this research.
Balinese Aromatherapy Product Design Element
In order to achieve the second objective of this research, product design
elements of both mixture composition and product container were identified from
each sample collected. The product samples collected was then evaluated
according to the new design concept generated by using the third questionnaire
(see Appendix 4). Thus, the result was analyzed using Relief method to obtain
which design elements are relevant to the new design concept generated.
Quantification Model of Balinese Aromatherapy Product
Thus, in order to accomplish the third research objective, firstly, the
personality test of respondent was evaluated using questionnaire as seen in
Appendix 4 (Littauer 1996). Afterward, quantification model of aromatherapy
mixture composition was generated as a result of analysis of the new design
concept and the product design elements. On the other hand, the level of Balinese
sense of the product was then evaluated by the experts using a questionnaire as
seen in Appendix 5. A quantification model of aromatherapy product container
was generated after fuzzification step was perform on both design elements and
product Balinese sense. Finally, the optimum product container design was
generated as a result of the relationship of new design concept and product design
elements with considered product Balinese sense for each customer personality
type.
Model obtained from both QTT1 and FQTT1 models would be produced
two main parameter, named category score and partial correlation coefficient. The
highest value on both category score and partial correlation coefficient were then
generated to obtain design support information on both aromatherapy mixture
composition and its container. The rule generated from design support information

7

would be evaluated using Fisher Exact Test in order to produce the productive
rules of new Balinese aromatherapy design for each customer personality type.

Data Analysis
Principal Component Analysis
Principal Component Analysis (PCA) is a powerful tool of feature extraction
which could reduce the dimensional data without losing much information
(Bouzalmat et al. 2014). In this research, PCA were used to reduce the dimension
of KW collected and followed by generating a new concept of product design
(Nagamachi 2011). To deploy PCA in this research, firstly, defines X as a matrix
of KW evaluation for p dimensions of KW which denotes as X = {x1, x2, …,
xp}Corr where xi € X. Then calculates the mean (μi) of X for n product sample (i =
1, 2,…, n) as in (1) which denotes as M = {μ1, μ2,…, μp}.

i 

1 n
xi
n
i 1

(1)

The next step is to calculate the subtract matrix between X and M as in (2) which
denotes as Y = {y1, y2,…, yp}.

yi  xi i

(2)

Afterward, in order to see the correlation between KW, compute the covariance of
matrix of Y which denotes as C with p x p dimensions as in (3).
C=

1
YT * Y
(n -1)

(3)

Generate eigenvalue and eigenvector of C as formulated in (4) where eigenvector
denotes as B = {b1, b2,…, bp} and eigenvalue denotes as λ = {a1, a2,…, ap}.
C*B=λ*B

(4)

According the result of eigenvalue λ, determine how many principal
component (PC) should be retained. Generally the PC which has eigenvalue above
1 were retained. Finally, Group the KW into their PC according to their loading
components which are generated from matrix eigenvector B and the new design
concept is then obtained from the interpretation of KW within PC obtained.
Relief
Relief method is a feature weight based method which detects those feature
which are statistically relevance to the target concept (Kira and Rendell 1992).
The key idea of Relief is to estimate attributes according to how well their values
distinguish among the instances that are near to each other (Kononenko et al.

8

1996). The equation of Relief is represented as in (5). In the equation, first,
defines ith feature (design element) as fi where F = {f1, f2,..., fm}, the weight of the
ith feature as wfi where W = {wf1, wf2,…,wfm}, and the number of each feature data
record for m dimensional features (i= 1, 2,…,m) and n sample size (t = 1, 2,…, n)
which denotes as xti where X = {xt1, xt2,…, xtm}.
(n1)

wfi 

  diff (x , y )
n

2

t 1 j (t 1)

t

j

(n 1)

(5)

Generally, Relief equation above runs using pairwise between data record (xt
and yj) where xt indicates an instance of feature data record and yj indicates an
instance neighbor of xt. In equation above, function diff (xt, yj)2 is calculated as in
(6), where rfi is a range unit of ith feature attribute to normalize the values of the
function into the interval [0, 1].

  x  y 2 
+ t j   (if xt and yj are the different class)
  rfi  


2
diff (xt , yj )  

2
  xt  yj  
(if xt and yj are the same class)


 rfi  


 

(6)

As in (6), if pairwise between xt and yj are in the same class, then the value
of function diff is negative. Otherwise, function diff will have a positive value
when xt and yj are in a different class. In this research, Relief method was used to
figure out the relevant design element of a product for the new Balinese
aromatherapy design concept.
Quantification Theory Type I
QTT-1 method considered as a qualitative and categorical multiple
regression analysis method, which allowed inclusion of independent variables that
are categorical and qualitative in nature, such as product design elements and
quantitative criterion variables within KE (Wei et al. 2011). In QTT-1 method, the
relational model associated with the measured score of experimental sample and
customer preferences is defined. Thus, the categorical multiple regression model
is defined as in (7):
E Ci

ys  ij xijs 
k

i 1 j 1

(7)

As in (7), ̂� expresses the predicted value of customer impression for the
sth product sample (s = 1, 2,…, n) on corresponding kth design concept. The index
of design element is denoted by i where i = (1, 2,…, E) and the number of design
element is denoted by E. The index of design category is denoted by j where j = (1,
2,…, Ci) and the number of category of the ith design element is denoted by Ci.

9

Additionally, � indicates the category score of the jth style within the ith design
element, � denotes the coefficient of the dummy variable which has binary
number value [0, 1] and � expresses a stochastic variable whose expectation value
E(�) = 0.

((Y-Y)(Y-Y))2
  r (Y,Y) 
(Y-Y)2 (Y-Y)2
2

2

(8)

The principle for quantification is to find the combination of category score
which maximizing the sample correlation coefficient (ρ) between observed value
̂) where �
̂ = { ̂ , ̂ , … , ̂� }.
(Y) where Y = {y1, y2,…, yn} and predicted value (�
The correlation coefficient between observed and predicted value is represented as
̅
̅ indicates the average of observed value and �
̂ indicates the
in (8) where �
average of predicted value. In this research, QTT-1 is proposed to figure out the
optimum linear combination of aromatherapy product mixture composition for
each personality type.
Fuzzy Quantification Theory Type 1
Fuzzy Quantification Theory Type 1 (FQTT-1) is a qualitative regression
analysis method that aims to analyze the relationship between qualitative
descriptive variables which are given in values on [0, 1] and numerical object
variables in the fuzzy groups given in the sample (Terano et al. 1987). FQTT-1
model as represented in (9) means a linear function of categories that best
expresses the structure of the data by minimizing the object variation given and its
error variance (Terano et al. 1987).

y()  aii()
K

(9)

i1

In model above, y(ω) expresses the objective function of ω th sample and μi
(ω) is the degree of response to ith qualitative category (i = 1, 2,…, K), which has
values on [0, 1]. The category weight (a) that minimizes the error variance is
given by the following equation (Terano et al. 1987) as in (10).
A = (XTGX)-1 XTGY

(10)

In equation (10), matrix A denotes a set of K category weights, where A =
{a1, a2,… , aK}. Matrix X contains groups of qualitative categories of n sample
given which denotes as X = {μi (1), μi (2),…, μi (n)}. Matrix G is a diagonal matrix
of fuzzy group B (μB (ω)) that expresses the fuzzy set of n sample given as
denotes in G = diag (μB (1), μB (2),…, μB (n)). Finally, the objective functions of n
sample given is contained in matrix Y which is defined as Y = {y (1), y (2),…, y
(n)}.
In order to figure out the influence of each category on object variable y
when changes in the other categories are fixed, fuzzy partial correlation
coefficients (fuzzy PCC) was generated in this model. To obtain the result of

10

fuzzy PCC, first calculate fuzzy means and fuzzy covariances. With considered
xi(ω) = aiμi(ω), fuzzy means for ith category ( ̅ ) and objective variable ( ̅) are
calculated as in (11) and (12). The next step in to calculate fuzzy covariances (σ)
as seen in (13), (14), and (15).

xi 

1 M n

xi ()B ()
N r1 1


(11)

y

1 M n

y()B ()
N r1 1


(12)

ij 

1 M n
(xi ()  xi )(xj ()  x j )B ()
N 
r 1 1

(13)

iy 

1 M n
(xi ()  xi )(y()  y)B ()
N 
r 1 1

(14)

yy 

1 M n
(y()  y)2 B ()

N r1 1

(15)

Using these covariances, matrix of fuzzy correlation coefficient is
generated which is denoted as R with i x j dimension where R = {r11 , r12 , …, rij ,

r1K , riy } and i = j = 1, 2, …, K. The formulation of fuzzy correlation coefficients rij
and riy are represented as in (16) and (17).

rij 

ij
ii jj

(16)

riy 

iy
iiyy

(17)

The invers of matrix R is then calculated which is denoted as R-1 = {r11 ,
r12 ,…, rij , r1K , riy }. Using this matrix R-1, The fuzzy PCC which is denoted as riy: 1,
2, …, i-1, i+1, …, K are defned as in (18).

riy:1,2,...,i1,i1,...,K 

riy
rii r yy

(18)

This partial correlation coefficient shows the influence of ith variable on the
objective variable when the other variables are fixed. Thus, in this research,
FQTT1 method was used to find the relationship between product design elements
and customer personality type for describing concept given.

11

Fisher Exact Test
In this research, Fisher exact test were used to assess the rule productivity of
design support information generated. In this method, the test directly conducted
to show whether the rule R: fi  Y is productive or not (Zaki and Meira 2013).
Let consider fi as a feature or design element generated where F = {f1, f2, …, fm}
and Y refers to customer preference which score above 4. The equation of p-value
calculation based on fisher exact test is represented as in (19).

p value 

(a b)! (c  d)! (a c)! (b  d)!
n! a! b! c! d!

(19)

In equation above, “a” denotes the value of appearance probability of both
fi and Y, “b” denotes the value of appearance probability of fi but not Y, “c”
denotes the value of appearance probability of Y but not fi, “d” denotes the value
of appearance probability of neither both fi nor Y in dataset given, and “n”
denotes the number of record within dataset. Thus, from the result of equation as
in (6), the rule is considered as a productive rule when the result of p-value is less
or equal than significance threshold given (α).

4 RESULT AND DISSCUSSION
Design Concept of Balinese Aromatherapy Product
Generating the design concept of aromatherapy product is necessary to
develop the product which suitable with customer good impression. In this
research, design concept was generated from the collection of customer
impressions related to Balinese aromatherapy product. PCA method was deployed
to analyze the impressions and to acquire the important information lies on
customer impression correspond to generate the design concept of Balinese
aromatherapy product. In this section, it is discussed about how to determine
aromatherapy product domain, and how to generate the design concept from the
collection of customer impression.
Product domain determination
Aromatherapy is defined by the Sense of Smell Institute (SSI) as the
therapeutic effects of aromas on physical conditions (such as menstrual disorders,
digestive problems, etc) as well as psychological conditions (such as chronic
depression) (Baser and Buchbauer 2010). There are virtually thousands of
aromatherapy products in pharmacies, high street shops, supermarkets, hair salons,
and beauty salons. The product are supposedly composed of essential oil (which
are usually perfumes) which has an effect on brain via inhalation, skin absorbsion,
or even directly via the nose (Baser and Buchbauer 2010).
Basically, the aroma of essential oil has particular effect on customer brain.
The process begin when the aroma was inhaled by customer, the molecule of the

12

aroma will bound by reseptors in olfactory bulb. These reseptors will send the
signals go through amigdala and limbic system in the brain. These parts of brain
have a big contribution related to human memory and emotion (Butje et al. 2008).
Olfactory stimulation causes immediate physiological changes in blood pressure,
muscle tention, pupil size, blink magnitude, skin temperature, skin blood flow,
and arousal states (Kuroda et al. 2005). Inhaled odors activate the release of
neurotransmitters such as serotonis, endorphins, and norepinephrine in the
hhypotalamic pituitary axis and modulate neuroreceptors in the immune system,
altering mood, reducing anxienty, and interrupting the stress response (Shannon
2002).
In this research, Balinese aromatherapy product domain was determined
based on expert recommendation. A set of questionnaire (see Appendix 1) was
given to three owners of Balinese aromatherapy product manufacture as the
experts in this research. The result in this phase showed that Balinese
aromatherapy product is an aromatherapy product which made from Bali
indigenous flower in Bali. Several Bali indigenous flower which are popular in
this business consist of frangipani, champaka, ylang-ylang, and sandalwood.
Additionaly, “the most potential Balinese aromatherapy product to develop”
was identified in this phase. The result of product domain identification which
represented in Appendix 6 showed that Balinese aromatherapy massage oil is the
most potential product to develop due to two main reasons, first, aromatherapy
product is closely related with spa which is identical with massage activity and
second, aromatherapy massage oil is a multifunction product which can be used as
a lotion, burn oil, bath oil, skin nutrition, and massage oil itself. Finally, based on
the result of this phase, frangipani, champaka, ylang-ylang, and sandalwood
massage oil was chosen as the object of the research.
Design concept generation
In this phase, twelve KW have been collected by means interviewing the
experts which illustrated Balinese aromatherapy product. These KW consisted of
“unique”, “attracting”, “exotic”, “natural”, “Bali product”, “luxurious”, “artistic”,
“bright”, “traditional”, “fashionable”, “eco-friendly”, and “transparent”. Then to
generate the new design concept, all the KW collected were evaluated by 30
customers of aromatherapy product. Likert scale (7 scale) were used in this
evaluation phase. The result of KW evaluation is represented in Appendix 7.
Afterward, the result of evaluation were analyzed using PCA method. The
result of PCA method by using R language is represented in Appendix 8. As
discussed previously in data analysis, there were two main PCA result in this
research consisted eigenvalue and loading factor. The result of eigenvalue of each
component is represented in Figure 2 and the result of loading factors of two first
components is represented in PCA variables factor map as seen in Figure 3.

13

Figure 2 The result of eigenvalue
Figure 2 showed the result of eigenvalue of each component. The larger
eigenvalue shows the more principal component represents the characteristics of
the original data. Based on PCA result in Figure 2, first principal component
which has eigenvalue above 1 (3.818) was generated to represent the information
of all the KW collected. The result showed that the first principal component has
79.51% contribution to describing the characteristic of the original data.

Figure 3 Variables factor map (PCA)
Then the KW were grouped into their principal component as seen in Figure
3. Based on the result of PCA, “Bali Traditional” (BT) was chosen to illustrate the
group of first principal component (“unique”, “attracting”, “exotic”, “natural”,
“Balinese product”, “luxurious”, “artistic”, “traditional”, “fashionable”, and “ecofriendly”) as a new design concept of Balinese aromatherapy product as the
answer of first objective in this research.

14

Balinese Aromatherapy Product Design Element
In KE research, product design element has a big contribution to create
customer attractiveness (Nagamachi and Lokman 2015). For instance, customers
will feel attracted when they see a new product and they will check the product
whether it impacts their feelings or not. The main idea is when the customer don’t
feel attracted and feel an impact when they touch the product, the opportunity to
get touch with customer is not developed and there no chance of the product being
purchased. This condition showed that product design elements is essential to
give the impression to customers directly (Muharam 2011).
In this research, the design of aromatherapy product was devided into two
parts consisted of its mixture composition and its container design. Each of them
has particular role on customer buying decision. In this research the aromatherapy
product design on both mixture composition and container were decomposed into
their design element and followed by classifying the product collected into its
design element. In this section, it is discussed about how to identify the product
design element on both mixture composition and product container and to
evaluate which elements are relevant to the design concept generation.
Design element of mixture composition
The identification of aromatherapy mixture composition was conducted on
the sample product collected. There are two main composition which compounded
in aromatherapy massage oil which consisted of essential oil and carrier oil. A few
drops of essential oil or oils chosen are always mixed with a carrier oil before
being applied to the skin for an aromatherapy massage (Baser and Buchbauer
2010). The exact dilution of the essential oils in the carrier oil is often
controversial and can be anything from 0.5% to 20% or more (Baser and
Buchbauer 2010).
Table 1 Identification of product mixture composition
Mixture composition
Type
1
2
3
4

Essential oil
(X1)
Frangipani
Champaka
Ylang-ylang
Sandalwood

Carrier oil
(X2)
Coconut oil
Olive oil
Soybean oil

As in Table 1, the identification result showed that there were four kind
essential oils (X1) identified that produced the aroma which closely related to
Balinese aromatherapy product consisted of frangipani, champaka, ylang-ylang
and sandalwood. These essential oil evaporates and has a concentrated aroma in
the product. In Bali, these typical aroma are commonly found in Balinese culture
which relates to its traditional ceremony or ritual.

15

The other result showed that there were three carrier oils (X2) compounded
in the product collected, consisted of coconut oil, olive oil, and soybean oil. All
the products used vegetable oil as its composition which represented the
traditional aspect of Bali indigenous. Carrier oil, also known as fixative oil or base
oil, compounds most of the liquid in massage oil. It is used to dilute (carrying)
essential oil and all provide a lubricant which many have a high vitamin E and A
content (Baser and Buchbauer 2010). After the mixture composition compounded
in each product was identified, the product collected were then classified based on
its design element, evaluated corresponds to BT concept by 30 respondent, and
finnaly analyzed using Relief method. The result of the design element
classification and evaluation are represented in Appendix 9.
For instance in Appendix 9, sample 1 has grouped into its design category as
follow: “X1=2” and “X2=3”. Refers to Table 1, these result showed that sample 1
has “essential oil: ylang-ylang” and “carrier oil: coconut oil”. The evaluation
score of Bali Traditional impression of sample 1 is 2.949. This result is the
average score of 30 respondents. Because the result has the score which is smaller
than 4, then the sample 1 is classified as NBT (Non Bali Traditional). On the other
hand, if the average score of Bali Traditional impression is larger than 4 which
showed in sample 2 (5.815), then sample 2 is classified as BT (Bali Traditional).

Figure 4 The Relief result of product mixture composition
All these data in Appendix 9 were analyzed using Relief and the result is
represented in Figure 4 and Appendix 11 (a). As seen in Figure 4 and Appendix
11(a), the result of Relief method represented that all design element of mixture
composition have shown the weight of attribute importance more than zero which
mean these two design elements are relevant related to new concept generated.
The result implies that designers should focus on their attention to both
aromatherapy essential oil and its carrier oil when the design objective of the
product is to achieve a desirable BT concept.
Design element of product container
The result of design element identification phase showed that from 25
product sample collected, there were 12 categories with 34 varians of design
elements identified using morphological analysis. The identification result is
represented in Table 2. Afterwards, the design element of all gathered products
were grouped into their categories, evaluated based on BT concept using Likert
scale (7 scale) and classified into two categories as seen in Appendix 10.
For instance in Appendix 10, sample 1 has grouped into its design category
as follow: “X1=3”, “X2=1”, “X3=2”, and so on. Refers to Table 2, these result
showed that sample 1 has “top shape: type 3”, “body shape: type 1”, “heigh and

16

width ratio: narrow”, and so on. The evaluation score of Bali T