Binus University Repository

Thoha, Nurianna

    
 
 
 
 
 
GCBSS © 2017 Global Academy of Training & Research (GATR) Enterprise. All rights reserved. 
Global Conference on Business and Social Science Series 
            conference homepage: www.gcbss.org 
             
 
Discovering the Relationship among Vrooms’ Motivational 
Factors Perceived by Generation Y 
(A Case Study in Corporate Service Organizations within the Greater 
Jakarta Area, Indonesia) 
 Dr. Nurianna Thoha, MBA1
*
, Daniel Adiputra Rasidi2 
1International Business/Faculty of Business BINA NUSANTARA University,  
JWC Campus – Jl. Hang Lekir I no 6, Senayan, 10270, Jakarta, Indonesia 
2 International Business/Faculty of Business BINA NUSANTARA University,  
JWC Campus – Jl. Hang Lekir I no 6, Senayan, 10270, Jakarta, Indonesia 
 
ABSTRACT 
 
Generation Y has started to dominate the workplace, including in Indonesia. It is expected that 75% of 
the workforce globally will consist of Generation Y by the year 2025. Therefore, it is urgent to look 
upon the different needs and motivations of this new generation, since they bring unique values, 
cultures, and motivations which are different from those of previous generations.  
Discovering Generation Y’s unique characteristics will allow a more in-depth and well-prepared 
motivation plan to be developed and implemented. Vroom’s Motivation Theory was examined and used 
as the fundamental theory for this research. 
This research applied quantitative studies. The pre-test questionnaire was analyzed through Cronbach’s 
Alpha and Kaiser–Meyer–Olkin as a reliability and validity test. The distributed questionnaires were 
then analyzed using T-test, Pearson Correlation, and Multiple Regression Stepwise. The final regression 
formula adapted was from stepwise regression.  
The research found that ‘salary’, ‘accessibility’, and ‘importance’ have the highest correlations for the 
‘expectancy’, ‘instrumentality’, and ‘valence’ variables respectively. 
The implications are a person’s motivation can be driven through salary increases, provide access to a 
higher-level manager, and create a job that is perceived to be important by him/her. Therefore, 
companies might motivate and retain Generation Y by incorporating these research findings when 
developing companies’ rules and policies.  
 
Type of Paper: Empirical 
Keywords: Expectancy; Generation Y; Instrumentality; Job-Motivation; Valence; Vroom’s Theory of 
Motivation. 
                                                          
*
 Corresponding author: 
                E-mail: nurianna@binus.edu 
                Affiliation: International Business/Faculty of Business, BINA NUSANTARA University 
 
Thoha and Rasidi 
2 | P a g e  
 
_________________________________________________________________________________ 
1. Introduction 
The rapid growth of Generation Y employees that are pouring into the labor market, create the 
need for attention with reference to the employees’ generational conditions. In 2015 Generation 
Y surpassed the number of Baby Boomers generation (Fry, 2015) and it is expected that 75% 
of workforce globally will consist of Generation Y by the year of 2025 (Schawbel, 2012).  
In Asia, the growth of the workforce is considered the highest in the world (Allen, Ahmed, 
Yip, & Switzer, 2014). It is estimated that around 90% of this workforce are living in 
developing or under-developed countries, in which they could not accommodate the large 
amount of labor (Sparreboom & Ernst, 2013). Furthermore, according to Sparreboom & Ernst 
(2013), there is an increasing percentage of Generation Y that started to get into a higher 
positions in the workplace from 12% to 87%.  
The purpose of this research is to discover and explore Vroom’s Motivation factors; such as 
valence; which is perceived important by Generation Y who work in corporate service 
organizations. The goal of this research is to allow managers to gain insights and formulate 
strategic planning and carefuly planned decisions towards the employees’ motivation in the 
workplace (Legault, 2002). 
2. Literature Review  
A generation cohort in social science is defined as a group of people within a delineated 
population that experience the same significant events in the time of their birth (Pilcher & Jane, 
2012). The individuals that have been grouped into a specific generation population may have 
developed common norms and moral values that are different from generations before or after 
them (Kowske, Rasch, & Wiley, 2010).  
2.1 Generation Y 
This generation is also called the Millennials; that is a group of individuals who were born 
between the years 1981 – 2000 (Nisen, 2013). This generation is known for their exposure to 
technology, multi-cultural environments, and multi-tasking work (Unsworth, 2008). They are 
also interested in business, computing, teaching, law, medicine, and hotels and tourism 
(Zemke, Raines, & Filipczak, 2000). This generation are slowly replacing the older generations 
in the workplace, and as pointed out by Park (2011) they value more meaningful and 
challenging work.  
Thoha and Rasidi 
3 | P a g e  
 
2.2 Vroom’s Expectancy Theory  
Vroom suggests that motivation is driven by behaviors from conscious choices among many 
alternatives that are available. He developed what is known now as ‘Vroom’s Motivation 
Theory’, which comprises ‘valence’, ‘instrumentality’, and ‘expectancy’.  
Valence refers to emotional orientation that people have in relation to outcomes or results. This 
orientation has two extreme points, which are positive and negative (Vroom, 1964). But, this 
does not eliminate the possibilities of other outcomes such as in the middle, or to a certain 
extent. Valence is a consolidation of three parameters that are ‘importance’, ‘attractiveness’, 
and ‘desirability’ (Thierry, 1996). 
The second term, ‘instrumentality’ explores the factors which shape and allow the expectation 
within a person. It is highly dependent on expectation which comes from past experiences, that 
implies certain efforts may result in certain outcomes (Brewis et al., 2007).  
The third term in Vroom’s model is called ‘expectancy’, that is simply the ‘action outcome’ 
association in a person’s mind. Several things can shape and change the ideals of expectancy, 
which has implications on the confidence of a person to achieve and accomplish his/her job. 
The main strength of Vroom’s theory of expectancy is that all the variables are not in control 
of managers. All three factors; expectancy, instrumentality, and valence, are connected to one 
another yet independent from external motives. This theory fits directly with Generation Y’s 
characteristics. This generation does want involvement that includes personal expressions.  
2.3 Hypothesis development 
Based on the discussion above, the research question is: “Which of the independent Vroom 
factors are perceived more significant on Valences’ variables by Generation Y workers in 
corporate service organizations within the Greater Jakarta area?”  
Thus, the hypotheses are: 
• H1: There is a significant relationship between independent Vroom factors towards 
importance perceived by Generation Y workers in corporate service organizations within 
the Greater Jakarta area. 
• H2: There is a significant relationship between independent Vroom factors towards 
attractiveness perceived by Generation Y workers in corporate service organizations 
within the Greater Jakarta area.  
• H3: There is a significant relationship between independent Vroom factors towards 
desirability perceived by Generation Y workers in corporate service organizations within 
the Greater Jakarta area.  
The framework of this research can be seen in Figure 1: 
Thoha and Rasidi 
4 | P a g e  
 
 
Figure 1. Research framework 
3. Research Methodology 
This research is explanatory research, where it is intended to discover and explain the 
relationship between variables in the research. The focus of this research is to investigate the 
variables within Vroom’s theory of motivation factors; that is valence, which may or may not 
be perceived as important by Generation Y. 
The pool for this research is Generation Y workers who are working in corporate service 
organizations. The company had to be a start – up, that had a maximum three years after 
founding, and under a venture or bigger corporate group. The research is done through 
purposive and quota-sampling methods. 
The total sample for this research are 108 respondents. These respondents represent four 
companies, with similar business practices, which still exist in the corporate service 
organization. The companies’ details are as follows; company A - offer services in a form of 
customer services towards individuals; company B is a cluster of companies that are under a 
venture; company C is a branding and activation company for new start – ups; company D is a 
start – up consultant in marketing and product engagement 
4. Results 
The data was analyzed using an Independent T-test and Multiple Regression. 
4.1 Regression 
Multiple Linear Regression was used to find the contribution and significance of an independent 
towards the dependent from other independent variables. 
Thoha and Rasidi 
5 | P a g e  
 
4.1.1 Mean Regression 
This section discusses the regression from Mean of Expectancy and Mean of Instrumentality 
towards Mean of Valence.  This give a bird’s-eye view of the research itself.  
Table 1. Mean regression coefficients 
Model Unstandardized 
Coefficients 
Standardized 
Coefficients 
t Sig. R 
Square 
B Std. Error Beta 
(Constant) 3.008 0.354  8.501 0  
0.037 Mean Expectancy 0.13 0.065 0.191 1.996 0.049 
Instrumentality Mean 0.009 0.083 0.011 0.11 0.912 
Table 1 shows that Mean Expectancy was largely significant in influencing the Mean of 
Valence. On the other hand, Mean Instrumentality had no significant influence on the Mean of 
Valence. In short, a quick depiction in terms of regression formula would be as follows: 
Vm = 3.008 + 0.130 Em + 0.009 Im   (1) 
A more in-depth look on the results of Model Summary shows that there was only 3.7% in 
Mean of Valence that can be explained in Mean of Expectancy and Instrumentality. 
4.1.2 Hypothesis 1 – Importance Regression 
Table 2. Importance (Valence) factor regression step wise coefficients 
Model Unstandardized 
Coefficients 
Standardized 
Coefficients 
t Sig. R 
Square 
B Std. Error Beta 
4 
(Constant) .103 .633  .162 .871 
. 216 
EXCha .290 .082 .315 3.543 .001 
INCom .300 .107 .257 2.790 .006 
INEmp .256 .086 .270 2.994 .003 
EXRel .124 .053 .215 2.364 .020 
 
Based on the result in table 2 and the four model, the new and Step Wise method of regression 
formula was: 
I = 0.103 + 0.290 EXCha + 0.300 INCom + 0.256 INEmp + 0.124 INRel   (2) 
In addition, this Step Wise method result showed that there were 21.6% of variability in 
importance that can be explained by this model alone. 
Thoha and Rasidi 
6 | P a g e  
 
4.1.3 Hypothesis 2 – Attractiveness Regression 
Table 3. Attractiveness (Valence) factor regression step wise coefficients 
Model Unstandardized 
Coefficients 
Standardized 
Coefficients 
t Sig. R 
Square 
B Std. Error Beta 
4 
(Constant) 4.225 .591  7.148 .000 
.281 
EXRec .290 .064 .384 4.518 .000 
INTra -.302 .076 -.349 -4.001 .000 
INWor -.315 .094 -.292 -3.339 .001 
EXCar .172 .082 .179 2.104 .038 
 
Taking from the table 3 and four final model, it would yield the following Regression Formula: 
A = 4.225 + 0.290 EXRec + 0.172 EXCar − 0.302 INTra − 0.315 INWor  (3) 
In addition, the model showed that 28.1% variability in attractiveness can be explained through 
these four most significant variables.  
4.1.4 Hypothesis 3 – Desirability Regression 
Table 4. Desirability (Valence) factor regression step wise coefficients 
Model Unstandardized 
Coefficients 
Standardized 
Coefficients 
t Sig. R 
Square 
B Std. Error Beta 
3 
(Constant) 3.753 .381  9.861 .000 
.192 
EXRec -.232 .071 -.290 -3.270 .001 
INTra .250 .081 .273 3.070 .003 
INRel -.159 .058 -.242 -2.734 .007 
 
From table four and three final model, it would yield the following regression model for 
desirability as: 
D = 3.753 − 0.232 EXRec + 0.250 INTra − 0.159 INRel  (4) 
Besides that, the model showed that 19.2% of variability in desirability can be explained by 
these new model (Recognition, Training, and Relationship with peers). 
 
 
Thoha and Rasidi 
7 | P a g e  
 
4.2 Correlations 
The simplified correlations are listed below as ‘expectancy’, ‘instrumentality’, and ‘valence’ 
variables are correlated to each mean correlation respectively. 
Table 5. Expectancy Correlation 
Expectancy 
Factors 
Correlation Sig. Rank 
Recognition 0.418 0.000 5th Rank 
Achievement 0.466 0.000 4th Rank 
Career 
Advancement 
0.606 0.000 2nd Rank 
Salary Increase 0.627 0.000 1st Rank 
Challenge 0.477 0.000 3rd Rank 
    
Instrumentality 
Factors 
Correlation Sig. Rank 
Employment 
Security 
0.060 0.537 5th Rank 
Compensation 0.394 0.000 4th Rank 
Working 
Conditions 
-0.25 0.794 6th Rank 
Relationship with 
Peer 
0.603 0.000 2nd Rank 
Accessibility to 
Higher Manager 
0.625 0.000 1st Rank 
Training 0.507 0.000 3rd Rank 
    
Valence Factors Correlation Sig. Rank 
Importance 0.437 0.537 1st Rank 
Attractiveness 0.406 0.000 2nd Rank 
Desirability 0.358 0.000 3rd Rank 
Table 5 shows the rank of the variables that belong to expectancy, instrumentality, and valence 
factors. Salary increase, accessibility to higher manager, and importance of the job are the most 
important variables perceived by Generation Y.  
5. Discussion 
Thoha and Rasidi 
8 | P a g e  
 
5.1 Importance factor 
The result indicate that importance is highly influenced by challenges in the expectancy area. 
Furthermore, it is also supported by employment security, relationship with peers, and 
compensation. It can be inferred that the importance of valence is motivated through different 
challenges that individuals are going to do while being compensated and supported with the 
instrumentality such as employment security and relationship with peers. 
5.2 Attractiveness factor 
In attractiveness, it shows that most variables that have a positive correlation come from 
expectancy, which can be implied as the attractiveness of a job or task to be done through 
motivation from the individual with the expectation of recognition and career advancement. 
5.3 Desirability factor 
The final factor of valence which was tested was desirability. In this factor, most of the results 
were negative while the positive correlation is only training, which comes from instrumentality. 
A quick implication is that an individual believes that the company could provide training in 
which he/she would learn to gain and use these skills in his/her tasks in the future. An important 
thing to note was the individuals tested had already got the job (s). 
6. Conclusion 
The Vroom’s motivational factors that are perceived important by Gen Y and the implications 
are as follows: for the ‘expectancy’ variables, salary is one of the highest correlations, followed 
by ‘career advancement’, and then ‘challenge’. The implication is a person’s expectation 
through ‘expectancy’ is most likely be driven through salary increases, because the person 
needs to fulfill his/her desires through monetization since money can fulfill his/her general 
living needs. 
For the variables of ‘instrumentality’, ‘accessibility’ is the highest, followed by ‘relationship’ 
and ‘training’. This finding can be interpreted as; a person will feel more motivated when 
he/she can have access to a higher-level manager. This depicts Generation Y as disliking 
hierarchical structures. Moreover, it also puts the onus on the employee to seek training and 
even mentorship from the higher levels. 
Companies and organization could also motivate and retain their employees by providing the 
significant variables of the respective ‘valence’ factors. In this case ‘importance’ is the highest; 
Thoha and Rasidi 
9 | P a g e  
 
followed by ‘attractiveness’, and ‘desirability’. It can be inferred that as a person will seek out 
the importance of the job first, then in return he/she will look for any other interests that may 
be valuable to him/her. This will then boost the attractiveness of the job and thus build up 
desirability towards the opportunity(ies) it may offer. These findings apply to Generation Y 
that work in corporate service organization in the Greater Jakarta area. It is hoped that 
companies will consider including measures in the companies’ rules and policies based on these 
findings.  
References 
Allen, A., Ahmed, S., Yip, G., & Switzer, C. (2014). The future-ready organization: reinventing work 
in ASEAN. 
Brewis, J., Contu, A., Grey, C., Kerfoot, D., Knights, D., Odih, P., & O'Doherty, D. (2007). 
Introducing Organizational Behavior and Management (2nd Edition ed.). (D. Knights, & H. 
Willmott, Eds.) Hampshire, United Kingdom: Cengage Learning. 
Fry, R. (2015, January 16). This year, Millennials will overtake Baby Boomers. Retrieved February 
16, 2016, from Pew Research Center: http://www.pewresearch.org/fact-tank/2015/01/16/this-
year-millennials-will-overtake-baby-boomers/ 
Kowske, B. J., Rasch, R., & Wiley, J. (2010). Millennials’(lack of) attitude problem: an empirical 
examination of generational effects on work attitudes. Journal of Business and Psychology, 
25(2), 265 - 279 . 
Legault, M. (2002). Bringing People Together: A study of generational diversity and organizational 
culture. Miami: University of Miami. 
Nisen, M. (2013, May 25). Definition of Generation Cohort. Retrieved March 26, 2016, from 
Business Insider: http://www.businessinsider.com/definition-of-generational-cohorts-2013-
5?IR=T&r=US&IR=T 
Park, J., & Gursoy, D. (2011). Generation Effect on the Relationship between Work Engagement, 
Satisfaction, and Turnover Intention among US Hotel Employees. Journal of Hospitality 
Business Management. 
Pilcher, & Jane. (2012, October 10). Mannheim's Sociology of Generation: An Undervalued Legacy. 
British Journal of Sociology, 45(3), 481 - 495. 
PWC. (2013). PwC’s NextGen: A global generational Study. California: University of Southern 
California. 
Schawbel, D. (2012, March 29). Career Strategies. Retrieved January 14, 2016, from TIME Inc.: 
http://business.time.com/2012/03/29/millennials-vs-baby- boomers-who-would-you-rather-
hire/  
Sparreboom, T., & Ernst, E. (2013). Global Employment Trends for Youth 2013 - A generation at risk. 
Youth Employement Program. United States: International Labor Organization. 
Thierry, H. (1996). Vroom's Expectancy Models and Work-Related Criteria: A Meta Analysis. 
Journal of Applied Psychology, 575. 
Unsworth, L. (2008). Four Generation Working Together. Retrieved March 1, 2016, from Ecu 
Education: http://www.ecu.edu/cs- lib/trc/upload/Four_Generations_Working_Together.pdf. 
Vroom, V. H. (1964). Work and Motivation. New York: Wiley. 
Zemke, R., Raines, C., & Filipczak, B. (2000). Managing the Clash of Veterans, Boomers, Xers, and 
Nexters in Your Workplace. Generational at Work. 


Discovering the Relationship among Vrooms’ Motivational Factors Perceived by  Generation Y  (A Case Study in Corporate Service Organizations within the Greater Jakarta Area, Indonesia)

Heriyati, Pantri

Form Reviewer "Franchising Performance from Franchisee Perspective: Case in Education Franchising in Indonesia"

Sasmoko, Sasmoko

Check JJA Prof Sasmoko _
Scopus 2018 - 21 - 9.
ITEI_Develop applications based
on android Teacher Engagement
Control of Health (TECH).pdf
by Arta Moro Sundjaja
Submission date: 13-Aug-2018 11:44AM (UTC+0700)
Submission ID: 623588408
File name: based_on_android_Teacher_Engagement_Control_of _Health_TECH.pdf  (1.5M)
Word count: 5043
Character count: 27991
















76%
SIMILARITY INDEX
33%
INTERNET SOURCES
62%
PUBLICATIONS
12%
STUDENT PAPERS
1 48%
2 12%
3 7%
4 3%
5 2%
6 1%
Check JJA Prof Sasmoko _ Scopus 2018 - 21 - 9. ITEI_Develop
applications based on android Teacher Engagement Control
of Health (TECH).pdf
ORIGINALITY REPORT
PRIMARY SOURCES
Sasmoko, S R Manalu, S A Widhoyoko, Y
Indrianti, Suparto. "Develop applications based
on android: Teacher Engagement Control of
Health (TECH)", Journal of Physics:
Conference Series, 2018
Publicat ion
icime.usu.ac.id
Internet  Source
china.iopscience.iop.org
Internet  Source
irep.iium.edu.my
Internet  Source
" The 2 International Conference on Computing
and Applied Informatics (ICCAI) 2017 ", Journal
of Physics: Conference Series, 2018
Publicat ion
ictromi.usu.ac.id
Internet  Source
7 1%
8 1%
9 1%
10 1%
Exclude quotes On
Exclude bibliography On
Exclude matches < 1%
Submitted to South Bank University
Student  Paper
bazybg.uek.krakow.pl
Internet  Source
investigadoresur.unirioja.es
Internet  Source
Submitted to Erasmus University of Rotterdam
Student  Paper


English

Similarity "Develop applications based on android: Teacher Engagement Control of Health (TECH)"

Girsang, Abba Suganda

Form Reviewer "Controlling System for Stock Raw Material for Production Planning and Inventory Control in a Pharmacy Company"

A Hybrid Cuckoo Search  and  K-Means for 
Clustering Problem 
 
 
Abba Suganda Girsang1 
1 Master in Computer Science, Bina Nusantara University,  
 Jakarta,  Indonesia  
Email : agirsang@binus.edu  
 
Ardian Yunanto2, Ayu Hidayah Aslamiah3 
2,3 School of  Computer Science, Bina Nusantara University, 
Jakarta,  Indonesia 
Email : 2ayunanto@binus.edu, 3aaslamia@binus.edu 
 
 
Abstract—Cuckoo search algorithm (CSA) is one of behavior 
algorithm which is effective to solve optimization problem 
including the clustering problem. Based on investigation, k-means 
is also effective to solve the clustering problem specially in fast 
convergence. This paper combines two algorithms, cuckoo search 
algorithm and k-means algorithm in clustering problem called 
FCSA.  Cuckoo search is used to build the robust initialization, 
while K-means is used to accelerate by building the solutions. The 
result confirms that FCSA's computational time in ten datasets is 
faster than the compared algorithm 
Keywords— Clustering, Cuckoo search algorithm (CSA), K-
means algorithm, Fast algorithm 
I. INTRODUCTION 
Some heuristic algorithms such as genetic algorithm, 
particle swarm optimization, ant colony optimization, bee 
colony optimization, cuckoo search (CS) algorithm and so forth 
are used to solve various optimization problem such as traveling 
salesman problem (TSP) [1] repairing  inconsistent matrix AHP  
[2][3], including solving clustering [4][5]. Clustering is a 
process to divide amount of data in the population based on the 
centroid point that constantly changing until it formed groups 
[6]. Many researchers have tried to develop the algorithms that 
more accurate or fast than the previous research for solving 
clustering, such as genetic algorithm [7], ant colony 
optimization [8], artificial bee colony optimization [9].   
One of swarm intelligent algorithm is used to solve some 
optimization problem is CS. Yang and Deb proposed firstly CS 
based  on  the  obligate  brood  parasitic  behavior of  some  
cuckoo  species  in  combination  with  the levy flight behavior 
of some birds and fruit flies  [10]. This algorithm was tried to 
solve some benchmark test functions. Then, many researchers 
continue this algorithm for solving some optimization problem 
such as solving traveling salesman problem (TSP) [11], 
clustering problem [12], multi objective [13], structural 
optimization problem [14], and so forth. CS and its modified 
are proposed by some researchers to solve clustering problem 
[12][15][16][17]. Unfortunately, in our best knowledge, no 
research is modified for CSA in fast model. Thus, the problem 
statement in this case is how to use the cuckoo search or its 
modified to solve the clustering problem with the efficient time.  
In original CSA for clustering, levy flight is used to generate 
random solution in each main iteration. Each main iterations is 
filled with pseudo iteration that calculate fitness from each new 
solution generated. The best fitness will be generated from the 
pseudo iteration to find out the best fitness of all cuckoo. 
Moreover, the centroid will be updated in first and last section 
of 1 pseudo iteration. In the last section, the algorithm uses 
random value to choose whether the iteration will generate new 
solution or not. This algorithm takes too much iteration which 
causes huge time process. It happens because each main 
iteration are filled with pseudo iteration that calculate fitness 
value from each new solutions generated in the pseudo 
iteration. Therefore, to optimize the redundancy of the process, 
pseudo iteration is much better to be discarded from the main 
iteration and replaced by K-means [12]. K-means is used for 
decreasing its computation time. Thus, the hybrid of cuckoo 
search and K-means are expected to solve the clustering 
problem with the short time. 
 
II. RELATED WORK 
Cuckoo search algorithm (CSA) is one of the latest nature-
inspired metaheuristic algorithm, which is developed in 2009 
by Xin-She Yang from Cambridge University and Suash Deb 
of C.V Raman College of Engineering [10]. Cuckoo search is 
based on a parasitism of some cuckoo species which chooses a 
nest where the host bird just laid its own egg. Commonly, the 
cuckoo eggs hatch earlier than the host eggs. Moreover, once 
the cuckoo hatches the new chick cuckoo, the cuckoo will evict 
the host egg by pushing the egg out of the nest. It makes 
increasing the portion of food for cuckoo chick which is 
provided by its host bird. The main step of CSA can be 
described as follows. First, some numerous host nests are 
generated. Each parasitic cuckoos lays one egg and dump its 
egg in randomly chosen nest. The best nest with high quality 
egg will be hatched in the nest and will be used as nest in next 
generation. 
The eggs laid by cuckoos which are discovered by the host 
bird with a probability pa ? [0, 1]. The host bird then abandons 
the cuckoo egg based on the value pa.   However the number of 
nest has to be fix. Therefore, substituting the abandon nest, the 
host bird creates a new nest so that the nest number is not 
reduced. This process is repeated until get the best solution.  
Figure 1 shows the Cuckoo search algorithm for clustering. 
Each nest represents a solution and a cuckoo egg represents a 
International Conference on Electrical Engineering and Computer Science (ICECOS) 2017
978-1-4799-7675-1/17/$31.00 ©2017 IEEE 120
new solution. The solution can be built by creating the centroids 
for each clusters. In the original CSA [10], the new solution of 
cuckoo is got by using levy flight approach. This new solution 
then compared to a nest solution which is selected randomly. 
The new solution is accepted if its solution is better than the 
nest’s solution. Then some of bad solutions (represents bad 
nest) are abandoned and some of new solutions will be built in 
order to retain the number solutions.  The best solution from all 
of the cuckoo and nest will be kept. The next process is 
conducted by repeating build the new solution of cuckoo using 
levy flight approach.  CSA has been applied in many areas of 
optimization and computational intelligence with promising 
efficiency. For instance, in computer science, CSA has superior 
performance over other algorithms for a range of continuous 
optimization problems, such as for Travelling Salesman 
Problem and clustering problem. 
III. THE PROPOSED METHOD 
The proposed method is divided into 2 main process: 
exploration (using CSA) and converging (using K-Means). 
First, initialization is conducted to adjust parameter used such 
as number nests, number generations, and so forth. The 
exploration is then done in step 2-6. Based on Yang's research 
[8], bird has an egg in some nests. An egg represents solution 
created which is built randomly. In clustering, the solution 
might be the centroids. Then a cuckoo is selected based on levy 
flight randomly. This cuckoo’s solutions is then compared to 
one of nest’s solution which is chosen randomly. The best one 
is kept, while the bad one is moved out. Using the coefficient 
fraction of pa, some worst of all solutions are discarded and 
then build the new ones to substitute the discarded solutions. 
The best of them is kept as initialization for k-means process. 
This k-means algorithm generates the new centroid as new 
solution. Then this solution is used for repeating the CSA 
process until the maximum iteration met.   
 
IV. EXPERIMENT RESULTS 
 
Fig. 1. Cuckoo search algorithm for clustering  [10] 
Matlab is used as a processing tools to develop and 
compile the proposed algorithm and comparison method. 
FCSA algorithm is then compared with original K-means 
algorithm for clustering problem [12,13]. The experiment 
International Conference on Electrical Engineering and Computer Science (ICECOS) 2017
121
method is implemented by using dataset retrieved from Uci 
Dataset, namely Iris, Wine, Yeast, Abalone, Breast cancer, 
Glass, E.coli, Haberman, Sonar and Parkinson [14]. Table 1 
describes the data type, the number of iteration and the number 
cuckoo nest 
  
Algorithm of FCSA 
1. Initialize the number of nests,  maximum generation for 
Levy flight 
//2-6 Exploration by CSA 
2. Generate random centroid for solutions corresponding of  
each the host egg nests. 
3. Generate some solutions (make centroid)  by some 
cuckoo egg. 
4. Get a cuckoo randomly by Levy flights evaluate its 
quality/fitness 
5. Choose a nest solution and comparing to the solution of 
step 4, keep the best one, and discard the other. 
6. Based on the fraction pa, some of solutions are discarded 
and some news are built, keep the best one solution.//This 
solution contains some centroids of solutions 
 
// 7-11 Converging by K- Means 
7. Each points are clustered to the respective centroids. 
8. Calculate the distance based on solution that already set 
by K-means algorithm. 
9. The smallest fitness will be stored as the best fitness 
value. 
10. The centroids are recalculated and storing as best solution 
11. Back to step 4 to generate the new solutions until 
generation for Levy flight  maximum achieved 
 
Dataset Iteration Cluster Cuckoo 
Nest 
Wine 300 3 10 
Iris 300 3 10 
Yeast 300 10 10 
Abalone 300 29 10 
Breast cancer 300 2 10 
Glass 300 7 10 
E.coli 300 8 10 
Haberman 300 2 10 
Sonar 300 2 10 
Parkinson 300 2 10 
 
The performance of FCSA is evaluated in two focus, quality 
and computation time. Quality means determining values taken 
from the sum of squared errors as described Eq. (1). 
 
??? ? ?? ? ???? ? ?????????????     (1) 
 
Quality is then observed by calculate the mean value and 
standard deviation from 10 experiment in the proposed 
methods. The result then compared by 10 experiments from K-
means method. The second focus is the computation process 
through generating the result. The computation proccess 
indicates the best computation from its 10 experiment, standard 
deviation and mean value. The the result will be compared with 
K-means method.  
Fig. 2 and 3 describe about the optimal fitness value reached 
between 50 iteration. From those figure, it can be inferred that 
the proposed method can achieve the optimal fitness less than 
50 iteration on those dataset. The proposed method get the 
lower fitness result than the comparison at first iteration, but 
achieve better result at next iteration.  Table 2 describes the 
calculation of mean and standard deviation quality.  
 
 
Fig. 2. Plot Fitness Wisconsin Diagnostic Breast Cancer. 
 
Table 2. Mean and standard deviation Quality of FCSA and 
K-means 
Dataset 
Name 
FCSA K-means 
Mean Std. Dev Mean Std. Dev 
Wine 2.4e+06 0 2.4e+06 8.04e+04 
Iris 78.94 0 85,33 20,21 
Yeast 468,78 94,06 489,92 154,71 
Abalone 27,73 0,62 27,93 0,74 
Breast 
Cancer 
7.8e+07 0 7.8e+07 0 
Glass 317.61 15.94 343,66 17,40 
E.coli 15,04 0,68 15,47 0,70 
Haberman 23404 17,18 30517 18,16 
Sonar 237,43 0,14 280,61 0,17 
Parkinson 3.4e+05 0 5.6e+05 0 
 
The result shows that from 10 dataset, almost all of dataset 
shows that our proposed method is more robust than compared 
Table 1. Data set and setting number cuckoo
 
International Conference on Electrical Engineering and Computer Science (ICECOS) 2017
122
method. Table 3 describes about the computational process 
from FCSA and K-means. From table 2 and 3, it can be 
concluded that FCSA have best processing fitness and 
computational time in almost all dataset 
 
Fig. 3. Plot Fitness Abalone. 
 
 
Table 3. Mean and standard deviation Computation Process of FCSA 
and K-means 
Dataset 
Name 
FCSA K-means 
Mean Std. Dev Mean 
Std. 
Dev 
Wine 0,014 0,003 0,19 0,02 
Iris 0,0075 0 0,16 0,002 
Yeast 0,08 0,06 1,25 0,11 
Abalone 1,22 0,13 4,71 0,78 
Breast 
Cancer 
0,04 0,005 0,41 0,02 
Glass 0,02 0 0,3 0,002 
E.coli 0,02 0,01 0,35 0,003 
Haberman 0.14 0 0,014 0 
Sonar 0,04 0 0,31 0,01 
Parkinson 0,02 0 0,023 0,01 
 
 
V. CONCLUSION 
 In this research, a fast algorithm is proposed to decrease 
the time process. The concept is to omit Levy flight feature and 
change it with K-means method to produce the best solution. 
Cuckoo Search will be used to avoid local optima that often 
happen in K-means weakness.  K-means is well known as 
solving of clustering problem and has ability to solve the 
clustering problem faster. The experiment result showed that 
the proposed method has ability to solve the clustering problem 
faster meanwhile the fitness value is almost equal with the 
compared method based on the same tools processing. The 
future work that can be taken from this research is to reduce the 
computation time of the proposed algorithm while at the same 
time optimize the quality of the results regardless of the 
  
REFERENCES 
 
[1] A. S. Girsang, C.-W. Tsai, and C.-S. Yang, “A Fast Bee 
Colony Optimization for Traveling Salesman Problem,” 
in Innovations in Bio-Inspired Computing and 
Applications (IBICA), 2012 Third International 
Conference on, 2012, pp. 7–12. 
[2] A. S. Girsang, C.-W. Tsai, and C.-S. Yang, “Rectifying 
the Inconsistent Fuzzy Preference Matrix in AHP Using a 
Multi-Objective BicriterionAnt,” Neural Process. Lett., 
vol. 44, no. 2, pp. 519–538, 2016. 
[3] A. S. Girsang, C.-W. Tsai, and C.-S. Yang, “Multi-
objective particle swarm optimization for repairing 
inconsistent comparison matrices." International Journal 
of Computers and Applica-tions 36.3, pp 101-109,  2014. 
 [4] I. B. Saida, K. Nadjet, and B. Omar, “A new algorithm for 
data clustering based on cuckoo search optimization,” in 
Genetic and Evolutionary Computing, Springer, 2014, pp. 
55–64. 
[5] S. Irvan, G. Robin Solala, and G. Abba Suganda, “An 
Adaptive Cat Swarm Optimization Based on Particle 
Swarm Optimization Approach (ACPSO) for Clustering,” 
Int. Rev. Comput. Softw., vol. 11, no. 1, pp. 20–26, 2016. 
[6] Z. Huang, “A Fast Clustering Algorithm to Cluster Very 
Large Categorical Data Sets in Data Mining.,” in DMKD, 
1997, p. 0. 
[7] J. C. Bezdek, S. Boggavarapu, L. O. Hall, and A. Bensaid, 
“Genetic algorithm guided clustering,” in Evolutionary 
Computation, 1994. IEEE World Congress on 
Computational Intelligence., Proceedings of the First 
IEEE Conference on, 1994, pp. 34–39. 
[8] P. S. Shelokar, V. K. Jayaraman, and B. D. Kulkarni, “An 
ant colony approach for clustering,” Anal. Chim. Acta, 
vol. 509, no. 2, pp. 187–195, 2004. 
[9] D. Karaboga and C. Ozturk, “A novel clustering approach: 
Artificial Bee Colony (ABC) algorithm,” Appl. Soft 
Comput., vol. 11, no. 1, pp. 652–657, 2011. 
[10] X.-S. Yang and S. Deb, “Cuckoo search via L{é}vy 
flights,” in Nature & Biologically Inspired Computing, 
2009. NaBIC 2009. World Congress on, 2009, pp. 210–
214. 
[11] X. Ouyang, Y. Zhou, Q. Luo, and H. Chen, “A novel 
discrete cuckoo search algorithm for spherical traveling 
salesman problem,” Appl. Math. Inf. Sci., vol. 7, no. 2, p. 
777, 2013. 
[12] R. Tang, S. Fong, X.-S. Yang, and S. Deb, “Integrating 
nature-inspired optimization algorithms to K-means 
clustering,” in Digital Information Management (ICDIM), 
2012 Seventh International Conference on, 2012, pp. 116–
International Conference on Electrical Engineering and Computer Science (ICECOS) 2017
123
123. 
[13] X.-S. Yang and S. Deb, “Multiobjective cuckoo search for 
design optimization,” Comput. Oper. Res., vol. 40, no. 6, 
pp. 1616–1624, 2013. 
[14] A. H. Gandomi, X.-S. Yang, and A. H. Alavi, “Cuckoo 
search algorithm: a metaheuristic approach to solve 
structural optimization problems,” Eng. Comput., vol. 29, 
no. 1, pp. 17–35, 2013. 
[15] N. M. Nawi, A. Khan, and M. Z. Rehman, “A new back-
propagation neural network optimized with cuckoo search 
algorithm,” in Computational Science and Its 
Applications--ICCSA 2013, Springer, 2013, pp. 413–426. 
[16] A. Hashmi, D. Gupta, Y. Upadhyay, and S. Goel, “Swarm 
intelligence based approach for data clustering,” Int. J. 
Innov. Res. Stud, vol. 2, pp. 572–589, 2013. 
[17] S. Arora and I. Chana, “A survey of clustering techniques 
for big data analysis,” in Confluence The Next Generation 
Information Technology Summit (Confluence), 2014 5th 
International Conference-, 2014, pp. 59–65. 
 
International Conference on Electrical Engineering and Computer Science (ICECOS) 2017
124


A hybrid cuckoo search and K-means for clustering problem

Manalu, Sonya Rapinta

Abbas, Bahtiar S.

Gaol, Ford L.

Lukas, Lukas

Trawinski, Bogdan

An Expert System to Assist with Early Detection of Schizophrenia

Warnars, Harco Leslie Hendric Spits

Sianipar, Nesti Fronika

Sanchez, Horacio Emilio Perez

Easy Understanding for Mining Discriminant Itemset with Emerging Patterns

Mulyadi, Martin Surya

Form Peer Review "Study of Income Tax Incentives in Creative Industry"

Yunus, Ulani

Form Reviewer "Ekpresi Religi Melalui Budaya Lokal di HKTY Ganjuran Yogyakarta"

Form Reviewer "Kepemimpinan di Indonesia Dalam Implementasi Dunia Perbankan Studi Kasus: Bank XXX Dan Bank Syariah ZZZ Jakarta"

Binus University

Not a member yet

Discovering the Relationship among Vrooms’ Motivational Factors Perceived by Generation Y (A Case Study in Corporate Service Organizations within the Greater Jakarta Area, Indonesia)