Desain Pelayanan Makanan Yang Menyenangkan Dalam Pesawat Berdasarkan Kepribadian Penumpang Menggunakan Hybrid Kansei Engineering.

PLEASURABLE IN-FLIGHT MEAL SERVICES DESIGN
BASED ON PASSENGER’S PERSONALITY TRAITS USING
HYBRID KANSEI ENGINEERING

HETY HANDAYANI HIDAYAT

GRADUATE SCHOOL
BOGOR AGRICULTURAL UNIVERSITY
BOGOR
2015

DECLARATION OF ORIGINALITY
AND COPYRIGHT TRANSFER*
Hereby, I declare that the thesis entitled Pleasurable In-flight Meal Services
Design Based on Passenger’s Personality Traits Using Hybrid Kansei Engineering
is my own work under supervision of Dr Eng Taufik Djatna, STP, MSi and Dr Ir
Hartrisari Hardjomidjojo, DEA. 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as unpublished papers are stated clearly in the text as well as in the references.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, November 2015
Hety Handayani Hidayat
Student ID F351130281

SUMMARY
HETY HANDAYANI HIDAYAT. Pleasurable In-flight Meal Services Design
Based on Passenger’s Personality Traits Using Hybrid Kansei Engineering.
Supervised by TAUFIK DJATNA and HARTRISARI HARDJOMIDJOJO.
In-flight meal service is one of the important point to judge an airline as
favorite to the passengers in the long trip. It is crucial to improved in-flight meal
services design that focused on pleasurable needs that highly was influenced by
psychological index. This index is known as personality traits. Hence, it should be
include passenger’s personality traits in designing services as pleasurable. Thus,
this work concentrated on formulating a pleasurable in-flight meal services design
by using hybrid Kansei engineering approach which combines Kansei engineering
and service system engineering. Kansei engineering used to excavated Kansei
word that related with in-flight meal service, whereas the service system
engineering contribute to identified and designed of costumized in-flight meal
service by using information technology.
This reseach done aims to identify the passenger’s personality traits, to

design in-flight meal service for each of the personality traits, and to evaluate the
model performances. Passengers Identification are carried by firstly collect Kansei
word from selected panelist interviewing, search Kansei word synonym by using
thesaurus dictionary and then have information retrieval of twitter and clustering
them by using Pillar K-means algorithm. To design an appropriate in-flight meal
service, it is neccesary to determine the design elements, collect the sample and
evaluation passenger’s preferences by questionnaire. They later became the basis
for the formulations synthesize the design by using quantification theory type 1
(QTT1). To assure that systems are implemented, the model performance will be
evaluated by t test.
The passenger’s personality traits identified from their tweet abaout in-flight
mela service on particular type namely neophobia, variety seeking selective, and
variety seeking. Variants formulation for each personality traits were designed for
different categories such as menu variant, menu information, appearances,
cordiality, originality, ordering method, and serving condition. In order to
simplify the user deployment, as result of pleasurable design corresponding to
each personality traits and presented in the dashboards. The evaluation by t test,
be discovered that the models have represented a real word. Requirement the
implementation of this model, it is required an integration fully with the current
booking and information costumer system that running online. As result obtained

the recommended design for user deployment to provide in-flight meal sevice that
corresponding to passenger’s personality traits.

Keywords: Hybrid Kansei Engineering, In-flight Meal Services, Personality
Traits, Pleasurable Design.

RINGKASAN
HETY HANDAYANI HIDAYAT. Desain Pelayanan Makanan yang
Menyenangkan dalam Pesawat Berdasarkan Kepribadian Penumpang
Menggunakan Hybrid Kansei Engineering. Di bawah bimbingan TAUFIK
DJATNA dan HARTRISARI HARDJOMIDJOJO.
Pelayanan makanan dalam pesawat adalah salah satu faktor penting bagi
penumpang dalam memilih maskapai yang disukai terutama untuk perjalanan
yang panjang. Hal ini menjadi penting untuk melakukan perubahan dalam desain
pelayanan makanan yang berfokus pada kebutuhan yang menyenangkan
(pleasurable) yang mayoritas dipengaruhi oleh indeks psikologis penumpang.
Indeks ini dikenal dengan istilah kepribadian. Oleh karena itu kepribadian perlu
dipertimbangkan dalam mendesain pelayanan tersebut. Formulasi desain pada
penelitian ini menggunakan Hybrid Kansei engineering yang mengkombinasikan
Kansei engineering dan service system engineering. Pendekatan Kansei

engineering digunakan untuk menggali kata-kata Kansei terkait pelayanan
makanan dalam pesawat, sedangkan service system engineering berkontribusi
dalam mengidentifikasi dan mendesain pelayanan makanan menggunakan
teknologi informasi.
Tujuan dari penelitian ini adalah mengidentifikasi kepribadian
penumpang, mendesaian pelayanan makanan yang sesuai, dan mengevaluasi
performansi model yang dihasilkan. Identifikasi penumpang dilakukan dengan
terlebih wawancara panelis yang terpilih untuk mengumpulkan kata Kansei,
selanjutnya adalah mencari sinonim kata Kansei dengan menggunakan Thesaurus
kamus online, serta melakukan pengambilan data dari Twitter dan klusterisasi
dengan agoritma Pillar K means. Untuk mendesain pelayanan yang sesuai, maka
dilakukan penetapan elemen desain, menggumpulkan sampel dan melakukan
evaluasi dengan menyebarkan kuesioner. Data ini kemudian menjadi dasar dalam
mensintetis model formulasi desain dengan menggunakan Quantification Theory
Type 1 (QTT1). Untuk memastikan bahwa sistem dalam diimplementasikan, maka
dilakukan evaluasi kinerja model dengan uji t untuk menguji realibilitas model.
Pada penelitian ini terdapat 3 kepribadian, yaitu neophobia, variety
seeking selective dan variety seeking yang teridentifikasi dari postingan tweet
yang disampaikan terkait pelayanan makanan dalam pesawat. Dalam mendensain
pelayanan makanan dalam pesawat, terdapat tujuh elemen desain terkait

kebutuhan yang menyenangkan (pleasurable) yakni variasi menu, metode
panyampaian informasi, penampilan, keramahan, asal menu, metode pemesanan,
dan keadaan penyajian. Dari hasil sintesis dengan QTT1, diketahui bahwa
pelayanan makanan untuk setiap kepribadian berbeda. Demikian juga dengan
elemen desain yang dipertimbangkan. Berdasarkan hasil uji t, diketahui bahwa
model dinyatakan telah merepresentasikan dunia nyata. Implementasi model ini
membutuhkan integrasi dengan sistem pemesanan dan informasi penumpang.
Penelitian ini merekomendasikan kepada penggunauntuk menyediakan pelayanan
berdasarkan kepribadian agar dapat meningkatkan preferensi penumpangnya.
Kata kunci: Desain yang menyenangkan, Hybrid Kansei Engineering,
Kepribadian, Pelayanan makanan dalam pesawat.

© 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 does not 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.


PLEASURABLE IN-FLIGHT MEAL SERVICES DESIGN
BASED ON PASSENGER’S PERSONALITY TRAITS USING
HYBRID KANSEI ENGINEERING

HETY HANDAYANI HIDAYAT

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

External examiner: Dr.Ir. Illah Sailah MS

PREFACE

I would like to thank Allah Subhanahu Wa Ta’ala for all His gifts and
favour to make research is successfully completed. The research is conducted
since July 2014, with the title of Pleasurable In-flight Meal Services Design Based
on Passenger’s Personality Traits Using Hybrid Kansei Engineering.
I would like to express my sincere gratitude to Dr Eng Taufik Djatna as
Chair of Advisory Committee for his support and encouragement during my study
in Bogor Agricultural University. I am very grateful to Dr.Ir.Hartrisari
Hardjomidjojo, DEA as Member of Advisory Committee for her advice and
supervision during the thesis work. I am very grateful to Prof Hanny Wijaya, Dr
Aji Hermawan, Prof Khaswar Syamsu, Ms Miranda Yasella and Ms Salma
Ibrahim as selected panelists who have given a lot of advices. I would like to
thank to my beloved family Sarif Hidayat (father) and Musilah (mother), Erny
Mulayasari Hidayat (sister), and Atep Nuryana Hidayat (brother) for their true and
endless love, for never failing patience and encouragement.
I would like to thank all lectures and staff of Agro-industrial Technology
Department, all of colleagues, especially my best friend in Laboratory Computer
of Agro-industrial Technology Department namely Novi Purnama Sari, Aditya
Ginantaka, IB Dharma Yoga, Rahmawati, Elfa Susanti, Elfira Febriyani, Nina
Hairiyah, M. Zaki Hadi, Azri Firwan, Rohmah, Wenny, Fajar Munichputranto,
Yudhis, Ikhsan, Imam and Denny, colleageus in Computer Sciences Department

namely Husnul, Riva, Puspa, Luki, Peter, Yudha, Siti, Heti and Ela), colleageues
in Dahlia namely Shierly, Dini, Iga, Maya, Anti, Rima, Mita, Nindya, Nana,
Agnes, and Fatma, colleagues in Agro-industrial comunity namely Ika Rezvani,
Dora Vitra Meizar, Priska Wisudawati, Ika Purwaning, Riri Mardaweni, Mustika
Zelvi, Felga Zulfia Rasdiana, Yosra Adi Putra Rully, Lely, Zulfa, Gilang, Wina
and to all of colleagues in Agro-industrial Technology 2013.
Last but not least, I would like to thank Directorate General of Highes
Education, The Ministry of Education and Culture (DIKTI) for BPPDN (Beasiswa
Pendidikan Pascasarjana Dalam Negeri) scholarship given.
I hope this research meet the requirement for Master Degree achievement
and its will be useful for the society.
Bogor, November 2015
Hety Handayani Hidayat

TABLE OF CONTENT
TABLE LIST

vii

FIGURE LIST


vii

APPENDIX LIST

viii

1 INTRODUCTION
Background
Problem Statement
Objective
Benefit
Boundaries

1
1
2
2
3
3


2 RELATED WORK

3

3 METHODOLOGY
Research Period and Location
Research Framework
Data Analysis
Identification Passenger’s Personality Traits
Formulation In-flight Meal Services Model Design Based on
Passenger's Personality Traits
Evaluation of Model Design

4
3
3
5
5
7

9

4 RESULT AND DISCUSSION
10
Passenger’s Personality Trait Identity
10
PleasurableIn-flight Meal Services Model Based on Passenger's Personality
Traits
15
Model Performances
27
Implementation Plan
28
Advantages and Disadvantages
29
5 CONCLUSION AND RECOMMENDATION
Conclusion
Recommendation

30
30
31

REFERENCES

31

APPENDIX

34

GLOSSARY

53

ABOUT THE AUTHOR

55

TABLE LIST
Kansei word from experts and literature review
Synonym of Kansei word
Data term frequency and tweet’s personality traits
4 Design elements
5 Sample identification
6 The evaluation passenger’s preferences
7 Design formulation for each personality traits
8 Top 3 design elements for each personality traits
9 Recommended design for each personality traits
10 Result of t test
11 Requirement data

1
2
3

11
12
13
17
19
21
22
23
27
27
28

FIGURE LIST
1

Research Framework
4
Optimized centroid algorithm
7
Examples of in-flight meal service
8
4 The example behaviour of personality traits
10
5 The best combination λand β in silhouette score
14
6 Distribution of tweet’s personality traits
15
7 Categories of Passenger's requirement
16
8 Histogram of Partial Corellation Coefficients
23
9 Selected dashboard as result of QTT1 analysis model for Neophobia (N) 24
10 Selected dashboard as result of QTT1 analysis model for Variety Seeking
Selective (VSS)
25
11 Selected dashboard as result of QTT1 analysis model for Variety Seeking
(VS)
26

2
3

APPENDIX LIST
List of questions to collecting Kansei words
34
Kansei word synonyms and symbols
37
Data acquisitions (data retrieve and cleaning data)
44
Clustering Pillar K-means Result
45
5 Passenger’s preference designs questionnaire
46
6 Respondent Profile
48
7 Passenger’s preference in-flight meal service designs questionnaire
(Neophobia Personality trait)
50
8 Passenger’s preference in-flight meal service designs questionnaire (Variety
Seeking Selective Personality trait)
51
9 Passenger’s preference in-flight meal service designs questionnaire (Variety
Seeking Personality trait)
52

1
2
3
4

1 INTRODUCTION
Background
Recent studies show that as air transportation industries rises significantly,
airlines have the main obligation to bring passengers from one place to their
destinations. However, along with the competitive climate growth, airlines are
also required to provide good quality services to their passengers starting from
before the flight, during the flight, and after the flight (Upadhyaya 2012). The
services provided are covering the ease of information and ticketing, the boarding
pass checking, the baggage provisions, the accuracy of schedule, comfort and
safety flight (Suki 2014; Jager and Zyl 2013; Archana and Subha 2012; Bahreini
et al. 2013; ACRP 2013; Jia EA, et al. 2012). For fulfilling the passenger’s
requirement, airlines also completing their services with a set aside food which is
known as in-flight meal (Jones 2004). In-flight meal services is one of the factors
on passenger’s list for choosing an airline which is covering food (i.e quality,
volume, menu variation, and appearance), pricing, cordiality of crew, getting
information, ordering method, and punctually.
Efforts to obtain optimum consumer satisfaction, each airlines is challenged
to fulfill passenger’s requirement. However, their requirement may be unlimited.
On the other hand, airlines have limited resources, especially budget. Therefore,
airlines should had examined passenger’s requirements deeply before they
designed excellent services. Referring to Jordan (2000) that the design of a service
or a product must consider functionality, usability and pleasurable aspects. But
according to ACRP (2013) in the case of in-flight meal services, functionality and
usability aspects had been met the current airline Standard Operational Procedure
(SOP). However, to improve customer satisfaction, airlines focus on pleasurable
aspect. In more details, pleasurable will be reviewed deeply of the passenger’s
psychological side (Jordan 2000). This psychological assessment becomes
important in designing favorable in-flight meal services. So, it becomes scope of
this research.
Designing service model is currently potentially performed with hybrid
Kansei engineering approach which combine the field of Kansei engineering and
service sciences. In contrast to other technique designs, Kansei engineering able to
deeply explore the explicit and the implicit factors of costumer’s want.
Customer’s psychological indexes are arrested as Kansei words (Nagamachi &
Lokman 2015). On the other hand, it is declared by Lopes and Ricardo (2013),
Service System Engineering (SSE) is a systems approach (consumer oriented
services) between the various stakeholders and resources that involved to design
customize and personalize services according to consumer’s requirement. SSE
contribute to deepen the design elements as well as information retrieval as a basic
clustering passenger’s personality traits in a short time or real time with IT-based.
According to Mak et al. (2012), in categorizing the factor that influenced
tourist food consumption was based on personality traits that only focused on the
interest rate to try new food. However, that study only discussed type of

2

personality traits. Therefore, it becomes challenge for this research to be able to
design in-flight meal service corresponding for each personality traits.
Kansei is defined as feelings and desires of each personality traits. In the
technological developments era, identifying personality traits are possible to done
through online and real time without having to ask directly to the person
concerned. One way is to use data from social media (Safko & Brake 2009).
Almost everyone have an account at the social media either Twitter, Facebook,
Google +, Path, Instagram, etc. In this study, the identification of passenger’s
personality traits be analyze based on their tweet (a opinion on twitter) in real
time. The related word frequency will be the basis for clustering. Twitter was
chosen as example because it has 284 million monthly active users, and 500
million tweets sent per day and then supports 35 languages with more than 40
million users (Duggan & Brenner 2013).
In order to enhance effectiveness, it is necessary to know the relationship
between personality traits and the design elements. This was conducted by the use
of Quantification Theory Type 1 (QTT1). According to Lai et al. (2006) QTT1 is
more effective and efficient. This QTT1 produce mathematical equation. The
resulting models are necessary to verify that the generated models are reliable.
Since the models are linear regression, they should be verify by t test (Pauole et
al. 2000). Based on the problems and challenges, the aims of this study is to
identify the passenger’s personality traits, to design in-flight meal service for each
of the personality traits, and to evaluate the model performances.

Problems Statement
It is great challenge for designers to formulate the service that fulfill the
passenger requirement and preference. In order to assist service designer, Hybrid
Kansei engineering was chosen as a method to design the costumized service.
Moreover passenger’s personality traits were involved to pursue the pleasurable
design which reflecting their own personality. Thus, there were several problems
related to in-flight meal service design based on Hybrid Kansei engineering based
on passenger’s personality trait, as follows:
1. How to identify passenger’s personality traits?
2. How to design pleasurable in-flight meal service for each of the passenger’s
personality traits?
3. How to evaluate the performances of proposed model?

Objectives
Based on described proposed problem above, there are three main objectives
formulated in this research as follows:
1. To identify the passenger’s personality traits
2. To formulate pleasurable in-flight meal service model for each of the
passenger’s personality traits
3. To evaluate the performances of proposed model

3

Benefits
The research is expected to give three benefits which consist of to provide
the information about new service design concept generation, service design
element which relevant to design concept, and quantification model of in-flight
meal service design. Advanced, the result of this research is expected to give the
recommendation about the best in-flight meal service that corresponding to each
passenger’s personality trait.
Boundaries
In order to concentrate for solution, following are the boundaries in this
research:
1. Meals were chosen as object of this research (except snack and beverages)
2. The design elements was focused on pleasurable needs of pasenggers
3. The input for identification of pasenger’s personality trait were data from
social media in text form. In this case, twitter was chosen as case study
4. The quantification model of in-flight meal service were generated from QTT1
analysis by using survey data

2 RELATED WORK
Study related to in-flight meal service was conducted since 2004. Jones
(2004) was studied about the operational of in-flight meal service and its
differences with food services. After 2004 until 2014, customer satisfaction has
been studied on the airline performance, included the in-flight meal service
quality. For example, Upadhyaya (2012) was studied about the customer
satisfaction measurement on the various airline at Arab Saudi. He postulate that
there was a gap between the costumer satisfaction and costumer need in many
parameter involved in-flight meal service. Also, Jia et al. (2012) was studied on
Malaysia Airline, then the service quality was studied on Indian Airline Archana
and Subha (2012). Then, Jager and Zyl (2013) studied in Malaysia and South
Africa Airline specially international passenger’s expectation. Bahreini et
al.(2013) was analysed the defferences performance between several airline. On
the otherhand, Suki (2014) was built the calculating model for service quality.
Kansei Engineering research are mostly used to designing product. But
Hartono (2011) was due to designing hotel service for Indonesian tourist. Then,
the end of 2014, Djatna and Hidayat (2014) was designed the in-flight meal
service by used real time key element but does not consider yet the pleasurable
needs and kansei of passengers.

4

3 METHOD
Research Period and Location
This research was conducted from July 2014 to April 2015. The samples of
in-flight meal services were collected in depth literature review. The survey
activities and data analysis were conducted at Computer Laboratory of Agroindustrial Technology, Bogor Agricultural University, Bogor.

Research Framework
The methodology was designed to solve the research problem. The research
framework detail was represented in Figure 1. In this research, generally, based on
interview and discussion with selected panelist, the Kansei word were obtained.
Its became an database to preprocess the posting content related to in-flight meal
service to identify the passengger’s preferences. Meanwhile in order to formulate
the pleasurable in-flight meal service model, the questionnaires were used to
acquire the data related to passenger’s preferences. Then, this models were evalute
by using t test.
Start
Define Personality traits
[literature review]

Collect the samples
[Market survey]

Collect the Kansei word
[experts & literature review
1st Questionner]

Identify of design
elements in the samples

T values

Search for synonims of Kansei word
[Thesaurus dictionary online]

Data samples

Determine T table
[significant level=5%]

Evaluate passenger’s peferences
[2nd Questionnaire]

Kansei word

Retrieve data in R languange
programming
[twitter]

Calculate T value element
designs in R

T tabel

Analyze correlation by QTT1
in R languange programming
T values > T table?

Preprocessing data
Categories value
& PCC
[Dashboard design]

Term frequency
per tweet

Valid design
Identify Personality Traits by
Pillar K-mean

Finish

Tweet’s
Personality Traits
clusters
Solution flow for 1st Objective

Solution flow for 2nd Objective

Figure 1 Research method flow chart

Solution flow for 3rd Objective

5

Referring to Neil and Noor (2011) that in-flight meal service given by an
airline differed between Low Cost Carrier (LCC) with legacy airlines. In this
research, in-flight meal service design devoted to legacy airline. Indefinitely that
the price attribute is unconsidered.
On the legacy airline, in-flight meal service given also distinguished by
class flight namely the economic, business and executive. Any of a class have
their own SOP. To be more specific on the sample collection and respondents, this
case focused on designing pleasurable in-flight meal service for economic class.
Economy class chosen as object class because the economy class has a more seat
portions, such as GA (Garuda Indonesia Airline) which has a total seats 96 consist
of the economic seats as many as 84 seats and the business ones only 12 seat (GA
2012).

Data Analysis
Identification of Passenger’s Personality Traits
There were several stages to identify passenger’s personality trait. Initially,
to define personality traits. Followed by, collecting the Kansei word related to inflight meal services by interviewing selected panelist and literature review. Then,
search the synonym of Kansei word by using Thesaurus dictionary online.
Thereafter, information retrieval from social media then data preprocessing.
Finally, clustering using Pillar K-means (Barakbah & Kiyoki 2009).
Definition of Personality traits
Personality traits needs to be defined to ease in understanding, identify and
clustering the data. A definition that will be used in this research are based on the
several literature review of previous research. Personality traits closely related to
someone’s psychological to response the problem that indicated with his or her
behaviour as early response. In this case only be limited to interest someone
would new foods
Collection of Kansei words
In this stage, Kansei words will collect by interviewing selected panelist.
The selected panelist were people who often enjoy various type of in-flight meal
service from various airline as well as having their interest against and understand
about him or his personality traits and the trends in choosing certain foods. To
further ease the selected panelist in collecting Kansei words, then used the
questionnaire on Appendix 1. This questionnaire contains the questions for
exploring the experiences, hopes and insights of selected panelists.
Searching synonyms of Kansei words
To be more effective in the next process, it is crucial to search the
synonyms of Kansei words. Searching synonyms aims to enrich database Kansei
words as identifier of personality traits. The synonyms was found out by using the
thesaurus online dictionary. A thesaurus was chosen as online dictionary because

6

it is one of the best online dictionary that containing synonym and antonym based
on the concept of meaning and rigorous with the other words (Caplan 2011).
Information retrieval from social media
Passenger’s personality traits were determined by using data or comment
from social media. Owing to the fact that, social media has become more popular
for people to express their ideology to public (Djatna & Hidayat 2014). Firstly,
determines the conditions of the data collection. Next, the data obtained need to
cleared redundancy and retweet to make sure there no duplication.
The major challenge in text mining is converting unstructured text into the
structured model. This must be done prior to doing any advanced analytics. The
possible steps of text preprocessing are the same for all text mining tasks, though
which processing steps are chosen depends on the task. The basic steps are as
follows (Miner et al. 2012):
1. Scope of the text to be processed was be choosen (documents, paragraphs, etc.).
2. Tokenize: Break text into discrete words called tokens.
3. Filter: Remove stopwords (“stopping”) or take wordlist.
4. Stem: Remove prefixes and suffixes to normalize words
5. sentence boundaries detection: Mark the ends of sentences.
6. Normalize case: Convert the text to either all lower or all upper case.
Clustering tweet personality trait by using Pillar K-means
In this purposed system, in order to cluster tweets apply Pillar K-means. In
pillar algorithm the distribution of dataset is similar to pillar of a building. The
initial centroid is located as furthers possible distance from the other. (Barakbah &
Kiyoki 2009). In other word, those k furthest object selected as initial centroid,
where k refer to cluster number to be observed.
In other word, those k furthest object selected as initial centroid, where k
refer to cluster number to be observed. Let T  ti | i  1,.., n be dataset, k be
number of clusters, C  ci | i  1,..., k be initial centroids, ST  T be identification
for T which already selected in the sequence of process, DM  ti | i  1,.., n be
accumulated distance metrics for each iteration and m be grand mean of X. The
proposed algorithm is described on Figure 2 (Barakbah & Kiyoki 2009).
After obtained the best initial centroid, the next step are following to Kmeans algorithm. K-means algorithm is an iterative technique that is used to
portion documents into K cluster. The basic algorithm are input k is the number of
clusters to get a set data of k clusters. The K-means steps are following:
a. k numbers of clusters were choosen to be determined
b. C from pillar solution as initial centroid was choosen
c. Each object to their closest cluster center were assigned using Euclidean
distance then compute new cluster center by calculating mean points

7

d. Step 2 and 3 were repeated until convergence is attained
Set C  , ST  , and DM  



Calculate D  dis T , m 
Set number of neighbors n min 
Assign d max  arg max  D 

.n
k

Set neighbors ndis   .d max
Set i  1 as counter to determine the i th initial centroids
DM  DM  D
Select   xarg max DM  as the candidate for the i th initial centroids
ST  ST 
Set D as the distance metric between T to 
Set no number of data points fulfilling D  nbdis
Assign DM     0

If no  n min
Assign D(ST )  0
C  CU 
i  i 1
If i  k , go back to step 7
Finish in which C is the solution as optimized initial centroid
Figure 2. Optimized centroid algorithm

Formulation of Pleasurable In-flight Meal Services Model Based on
Personality Traits
After discovered passenger’s personality traits,it is crucial to consider them
in designing in-flight meal services. To design pleasurable in-flight meal services
using hybrid Kansei engineering, there were several stage. Frist, collect the inflight meal services sample. Next, determine design elements based on service
science perspective. Then, evaluate passenger’s preferences. And the last is
synthesis design formulation using QTT1.
The following steps are quantification theory type 1 (Lai et al. 2006)
processes on how to:
1. The in-flight meal services attributes were determined (Xn) (n=1,2,.....,7)
2. the categories i of in-flight meal services attributes (Xni) are defined. For
example, in this case for menu variant there are 2 categories i.e 2 options (X11)
and vegetarian, moslem, and kosher (X12)
3. The samples are classified based on their attribute categories
4. The passenger’s preferences are evaluated about samples
5. The in-flight meal services are formulated by using QTT1 on R programming

8

Collecting in-flight meal services design samples
In-flight meal services design samples will obtain by market survey from
domestics and international flight with all of classes (economics, business, and
executive classes). Figure 3 below described in-flight meal service samples.

Figure 3. Example of in-flight meal service
Determining the design elements
Design elements should determine to identifying samples and to focusing
formulation in-flight meal service design .The identification of design elements
and their own types will be done by literature review on pleasurable needs. The
design elements were seen based on the service system engineering perspective in
hybrid Kansei engineering.
Evaluating the passenger’s preferences
In this stage, the evaluation created passenger’s preferences conducted by
survey. Hence, the selected respondent will answer the questionnaire. The
questionnaire was consist of in-flight meal samples. The preferences levels were
5-point Linkert scale, started from 1 that refer to “strong dislike” and 5 that mean
to “strong like”. The samples obtained from previous stage. The questionnaire is
presented in Appendix 5. The questionnaire was divided into two main parts is the
first part to know the number of respondents enjoyed the experience of in-flight
service meal service and his or her personality trait.
Synthesizing model formulation by Quantification Theory Type 1 (QTT1)
Formulating model design was adopted on the application of Kansei
Engineering to determine the relationship between design concept and design
elements by using Quantification Theory Type 1 (QTT1) method (Schütte et al.
2004). In this case, the design concept in standard Kansei engineering were
modified as personality traits. Infinitely, the distinguish stage between hybrid

9

Kansei engineering and standard Kansei engineering are in determining the design
concept.
Referring to Hui et al. (2009), the QTT1 define as a method of qualitative
and categorical multiple regression analysis allowed inclusion of independent
variables that are categorical. By using QTT1 synthesis of service will be more
powerful. The model formulation as follows (Lai et al. 2006):
^k

E

Ci

y s   xi. j ijs  

(2)

i 1 j 1

^k

Where y s is the predicted value of standard variable for the sth product sample
{s1, s2, ..,s15}, i on the kth personality traits {k1, k2, k3}, i is design elements index,
E is number of design elements, j is categories index, Ci is the number of
category of the ith design element { i =1,2,..,7},  is stochastic variable whose
expectation value E ( )=0, xi. j is the category score of the jth style within the ith
design element, and  ijs is dummy variable coefficient.

Evaluation of Model Performance
Evaluation means to ensure that the model has been represented in the real
world. In this research, the model was verified the reliability with t test. There
were several stages to evaluate model performance. Firstly, calculate t value. Next
stage is determine t table. Then compare t value and t table. Here is more detail to
evaluate the model performances:
Calculating t value
The model performance was verified the reliability with t test of PCC
(Partial Correlation Coefficient) values from the result of Quantification Theory
Type 1. The following questions detailed for calculating t value as follows (Hui et
al. 2009):
s  E 1
tvalue  r
(3)
1 r2
Where tvalue is reliability result, r is Partial Correlation Coefficient (PCC) value,
s is ordinal number of sample, E is ordinal number of design elements.
Determining t table
For determining t table, we used Table t with the  significance level,
(  = 0.05) used is 0.05. The significance level, also denoted as alpha or α, is the
probability of rejecting the null hypothesis when it is true (Johnson 1999). In this
case, a significance level of 0.05 indicates a 5% risk of concluding that a
difference exists when there is no actual difference.

10

Comparing t value and t table
After discovering t , in this stage, t value will be compare with t table.
Design element declared as reliable, if:
tvalue  t
(4)

4 RESULT AND DISSCUSSION
Passenger’s Personality Trait Identity
From our results we discovered that personality traits are psychological
factors influencing consumption patterns. This factor leads to passenger
characteristics that influence their consumption behavior. Kittler and Sucher
(2008) mentioned that there is a close relationship between food and the
personality traits. This includes the food selection, the presentation, the equipment
used to how to eat it. This is in accordance with the terms that are familiar in the
world i.e. “You are what you eat.” Personality traits is one of the things that
affects consumer wishes and judgment about food service.
The personality traits were divided into theree namely (a) neophobia, (b)
variety seeking selective, and (c) variety seeking as seen in Figure 4. According to
Mak et al. (2012), they postulate 2 types of personality traits are neophobia and
variety seeking. In they studied, define (a) neophobia as a people who are
reluctant to try new foods. Secondly, (b) variety seeking tends affect a personal
food choices. Variety seeking is the term used for the personality of the person
who likes looking for something that is diverse (diversity) and different as a good
choice in service or food. This type has the flexibility to adopt the food they
consume. In the development, it has been added by Djatna and Hidayat (2014), for
a consideration of some groups of people who have an interest but they were
constrained in certain allergies or limit certain types of food. The group is called
as variety seeking selective (c).

(a)

(b)

11

(c)
Figure 4. The example behaviour of personality traits
(a) Neophobia, (b) Variety Seeking Selective, and (c) Variety Seeking
Therefore, in this case their psychological captured from the words spoken
in their opinion on Twitter (tweet). The research began by interviewing 5 selected
panelist and literature review for collecting Kansei words. Selected panelist are
people who have often enjoy in-flight meal services of various international
airlines, has an interest about in-flight meal service and understand about food
personality traits. To assist selected panelist in collecting Kansei words, in
addition to interview, they also asked to fill the questionnaires on Appendix 1. As
a result, 57 Kansei words about in-flight meal services are obtained, as seen in
Table 1 below.
Table 1 Kansei word from experts and literature review
Personality traits
Kansei word
difficult, common, familiar, reject ,definite, colorless, local,
Neophobia (N)
satisfying, popular, single, old, tasty, same, similar, traditional
acceptable, agreable, healthy, homesuitable, delicious, deluxe,
Variety Seeking
fresh, juicy, friendly, kindly, halal, helpful, intelligent,
Selective (VSS)
nutricious, skill, thoyib, clean, selection, vegetarian, perfect
choices,colorful, attractive, new, cheap, expensive, difficult,
Variety Seeking flavorful, different, delectable, unique, brave, flexible,
(VS)
greeting, impressive, interest, unfamiliar, unpopular, special,
option, memorable, extraordinary, modern
Source: Experts survey (2014); Martin (2001); Mak et al. (2012); Ratner &
Kann (2002)

The Kansei words on Table 1 become the word identifier in filtering and
clustering step to identify passenger’s personality traits. Furthermore, the
synonyms of 57 Kansei words were searched to better accommodate to filter the
tweet contain. Based on thesaurus (online dictionary), as result, the synonyms
obtained are 1113 synonyms of Kansei words that shown on Table 2.

12

Table 2 Synonym of Kansei words
Synonyms

Words
limited, customary,
disesteemed, remarkable, challenging,
demanding, laborious, painful, formidable, galling, gargantuan,
hard-won, herculean, immense, intricate, irritating, labored,
operose, prohibitive, frequent, ..................................................,
coincident, concurrent, contempt, restored (Thesaurus.com)

Having Kansei word synonyms, next steps is data retrieval by using query
condition in R language. Tweet were taken that containing the words inflight
catering, inflight dining, inflight food, flight meal, airline meal, airline catering”
as condition with a sample of 1474 tweets updated with English language during
in the years of January 2014 to April 2015. The keywords election are based on
the other terms that often used to refer in in-flight meal services. It is number of
tweet which reflect of 1% of allowable access from total tweets in the worlds that
fulfill the conditions. This access is free of charge and legal from twitter users
who have API key .The API key is code unique which will be needed to
synchronizing blog or a web by social network account belonging to the user.
The basic assumption of which are applied in the taking of the tweets is all
the tweets that fulfilling a condition that has been set will come from passengers
or people who have been enjoying in-flight meal services from the certain airline.
Because account user that posting the tweet must have certain interest against inflight meal services. In addition, the tweets taken is not limited by an airline and
regions.
Tweets that have been taken then be removed the data redundancy and
spam. It turns out there were only 1065 of 1474 tweets that capable to process.
More detail about information retrieval was shown in Appendix 3. The amount of
data obtained for seventh the keyword are different. This indicates the level of
popularity keyword as a term used to denote in-flight meal service. Then the data
did preprocessing that includes tokenizing, filtering the Kansei word synonyms
that has been accumulated and stemming to find basic words. The Kansei word
synonym and notation are shown on Appendix 2. Synonym words should be
notated, to ensure the number of synonyms and to ease column labeling in output
produced. Output from this data preprocessing is term frequency matrix. There are
1113 synonyms of Kansei words which denoted as Kw1113 as a representation
that data term frequency matrix there will be consist of 1113 column.
After term frequency matrix obtained then subsequently used as the basic for
clustering. Clustering method chosen because the dataset does not have a label or
target class. As for the method used by using an algorithm optimized Pillar Kmean (Barakbah & Kiyoki 2009). The result of clustering seen in the Table 3.

13

No
Tweet
1
2
3
4
5
6
7
8
9
10
11
12
13
.
.
1067

Table 3 Data term frequency and tweet’s personality traits
Kansei words synonyms
Personality
traits
Kw1
Kw2
Kw3
Kw4
Kw1113
0
0
0
0
.
0
VSS
0
0
0
0
.
0
VSS
0
0
0
0
.
0
VS
0
0
0
0
.
0
VSS
0
0
0
0
.
0
N
0
0
0
0
.
0
VS
0
0
0
0
.
0
VS
0
0
0
1
.
0
VS
0
0
0
0
.
0
VSS
0
0
0
0
.
0
VS
0
0
0
0
.
0
VS
0
0
0
0
.
0
VSS
0
1
0
0
.
0
VS
.
.
.
.
.
.
.
.
.
.
.
.
.
.
0
0
1
0
.
0
VS

Data term frequency on Table 3, the Kansei word synonym (the notation
were shown on Appendix 2) calculation only consist of zero (0) and one (1). The
meaning of zero (0) are refer to that tweet does not content the Kansei word.
Otherwise, one (1) refer to Kansei word in tweet content only spoken once. Table
3 above were be an input in the clustering process by using Pillar K-means and
personality traits is the output produced. The clustering process are used the
algorithm that shown on Figure 2.
In this case, the optimum numbers of cluster were not sought because it was
determined before that there were 3 type peronality traits. Pillar K-means was
chosen because more accurate than standard K means. Because the pillar was
maximizing the distance between centroid and minimizing the scope area value of
cluster. The output of clustering was shown in Appendix 4. As a result, that
number of cluster is three ( k  3 ) that represents the kind of personality traits,
namely N (Neophobia), VSS (Variety Seeking Selective) and VS (Variety
Seeking). In addition also featured members of each clusters namely tweets post
numbers. The best cluster solution is only selected if it has no empty cluster and if
it has no negative average silhouette score as shown in Figure 5.

14

lamda: 0.1
beta : 0.2
Figure 5. The best combination λ and β in silhouette score
Before the centroid of each clusters are determined, the optimal value of α
and β must be determined by trial and error. The values of λ and β play significant
role in silhouette score. Silhouette function was to understand how good an object
is placed in a cluster (Barakbah and Kiyoki 2009).
The β (beta) value shows the gap or distance between centroid clusters and λ
(lamda) value refer to the range or radius each cluster. Hence, the basic principles
in Pillar K-mean to optimize initial centroid is to maximize beta value and
minimize alpha value. Based on calculation, the best cluster solution were λ = 0.1
and β = 0.2 because it had the highest silhouette score (s = 0.833938272).
To recap, the type of passenger’s personality traits is identified from a tweet
posting that related with in-flight meal service. From observed dataset, it is found
the passenger’s personality traits distribution as shown in Figure 6.

15

Figure 6. Distribution tweet’s personality traits
From Figure 6 above, known that the majority distribution of the tweets
personality trait is variety seeking with 71% or consist of 755 tweet posting. Then,
distribution of neophobia and variety seeking selective nearly the same namely 15
% or 154 tweet posting and 14 % or 153 tweet posting.

Pleasurable In-flight Meal Service Model Based on Personality Traits
In-flight meal service is an additional services provided by an airline to
passenger with set aside food during their journey, in-flight meal service is meant
to increase of overall passenger satisfaction. Lupiyoadi and Hamdani (2008)
stated that service in an activity that have occurred from interaction with a person
or a machine that which produces customer satisfaction .Although services are
intangible things, but basically just as product, services also respond to its
consumer acceptance. Hence to design the excellent in-flight meal services, the
airlines should to know the specific of passenger’s requirement.
According to Jordan (2000) that passenger’s requirement dirived into three
hierarchy level needs that namely level 1 is functionality needs, level 2 is usability
needs and level 3 is pleasurable needs. Then, Figure 7 below shown categories of
passenger’s requirement on in-flight meal services.

16

Figure 7. Categories of passenger’s requirement
(Adopted from Jordan 2000 and Perone et al. 2005)
Clearly, in the figure 7 shown the categories of pasenger’s requirement
about in-flight meal service. The service will be useless if it does not contain
appropriate functionality. In-flight meal service cannot be usable if it does not
contain the functions necessary to perform the tasks for which it is intended. If inflight meal service does not have the right functionality, set aside food with fast
and accurate, will cause dissatisfaction.
Having appropriate functionality is a prerequisite of usability, but it does
not guarantee usability. The usablity of in-flight meal service is comitement and
consistency services; providing the food safety; and acceptable costing. In other
words, passenger want to have in-flight meal service that provide the nutritious
and healthy food, with enough volume that serve in the right time when they
hungry.
Having become used to usable services included in-flight meal service, it
seems inevitable that people will soon want services that make passangers feel
good about who they are; services that bring not only functional benefits but also
emotional ones. In pleasurable needs, passenger want an in-flight meal service
that costumize, impress, memorable and enjoyable. Its level related to Kansei
concept that consider the psychological aspect to designing services. In addition,
based on ACRP document (2013) discovered that in airline’s current SOP
(Standard Operational Procedure) only fulfilling the level of functionality needs
and usability needs. But the third hierarchy level, pleasurable needs, are has not
been fulfilled. Hence, in this work focused on pleasurable needs.
Knowing attributes of in-flight meal service is an important step that
should be done if we will formulate its design. In this case, discussing 7 design
elements on scope of our problem. For more details, design elements and the type
were described in the Table 4.

17

Table 4 Design elements
Type Menu Variant

Menu
Information
(X2)

Appearance

(i)

(X1)

1

2 Options

Limited

Unique

Greeting

2

Kosher,
Muslim, And
Vegetable
Menu

Oral

Standard

More
Information

3

(X3)

Cordiality
(X4)

Written

4

Originality
(X5)
Origin
Departure

Ordering
method
(X6)
Package

Destination

Buy On Board
(Bob)

Not Specified
Determine Their
Own

Prebooking
BoB&
Prebooking

Serving
Condition
(X7)
Hot
or
Warm
Cool

Not only about meals but also about how these meals are served. The
attributes about meals are including:
a. Menu variant (X1)
Menu variant (X1) is how much an option granted to passengers in
choosing the food. It has 2 categories which usual be on the market that is 2
options (X1.1) such as rendang or fried rices; Kosher, Moslem, and vegetable
(X1.2). Kosher refer to food for person that have interdiction. Moslem that
meals non alcohols and insurable as halal food; and vegetable is meals non an
animal protein.
b. Appearances (X3)
Presentation of our meals with appetite for its consumption. So, airline
business must pay attention about appearance. This attribute have 2 categories
namely unique (X3.1) such as using banana’s leaf or another materials that refer
to local wisdom; and standard (X3.2) such as using the dishes standard.
c. Originality (X5)
In this case, originality that mean place where are the meals come from.
Based on point of departure, originality only derived as 4 categories namely
origin depature (X5.1), destination (X5.2), not specified or random (X5.3), and
determine their own (X5.4).
The attributes about service are consist of:
a. Menu Information (X2)
Menu information is how airlines explain menu choices offered . Its design
elemen derived as limited (X2.1) that mean only list of menu; oral (X2.2) is the
steward give more detail about menu such as the material, processes, etc; and
written (X2.3) is there are information detail in list of menu.
b. Cordiality (X4)
In this case, cordiality attributte about attitude of flight stewardess when
they offer the menu. They only greeting (X4.1) or they give more information
(X4.2).
c. Ordering method (X6)

18

Ordering method derived into 4 type. Package (X6.1) mean that the meal
are included in ticket and passengers does limited to choice the menu. Buy On
Board or BOB (X6.2) is the method that the passengers determine their menu
during the flight. Prebooking (X6.3) mean that the passengers should be select
their menu when they reservesed the ticket. Prebooking and BOB (X6.4), this is
more flexible because the passengers can selected the menu before flight or
during the flight.
d. Serving condition (X7)
Serving condition showing how conditions food when served to
passengers.This design elemen derived as 2 categories, namely hot or warm
condition (X7.1) and cool condition (X7.2).
After determine the elemen designs, the next stage is to collect the in-flight
meal service samples. From the results of the market survey, there were 15
samples of in-flight meal services. The selected sampels are in-flight meal
services type that provided from the airline in the world that were consist of
domestic and international flights. In this case, there were several airline that was