Customer Clustering Using The K-Means Clustering Algorithm in Social Media Users in

Indonesia

Authors : Fe

Abstract

Social media is the number one aspect in the world because it can contact countries regarding

communication, and shopping and easily reach access news entry quickly between countries.

As a strategy for the action plan in each social media platform, this research aims to be able to

identify the most existing social media users, based on the content they like from each of

these social media platforms, and be able to categorize content and targets. from segmentation

on each social media platform using the K-Means Clustering algorithm. There are ten

variables, age, gender, income, marital status, home location, job position, and interest in

content from social media. The application of this research uses SPSS V22 which assists in

data processing. The total population in this study was 400 participants. In this case 3 clusters

have each persona. The first cluster is Tik Tok, the second cluster is Instagram and the third

cluster is Twitter.

INTRODUCTION

The development of media technology is very rapid as well as with the advancement

of technologies that are increasingly developing both in Indonesia and throughout the

world. In this case, in particular, the country of Indonesia has experienced an increase in the

use of Social Media which is accessed from mobile phones as well as various kinds of

usage processes as we can see as well as from various sources which state that Indonesia
has experienced growth in the use of social media which places Indonesia currently in third

place with internet users. largest after China. In this case social media is consumed by

various groups in Indonesian society such as students, researchers, students and the general

public (Boediman et al., 2021).

Access to information and communication technology is now also not limited to the

younger generation, because the older generation and even children also use and feel the

impact of advances in information technology. Advances in information and

communication technology are used to meet various needs, starting from simply seeking

information, buying and selling online (e-commerce), to accessing social media. The high

intensity of use and utilization of social media nowadays also influences people's habits and

consumption of information (Li et al., 2021). Not only used for fun, social media also

provides easy access to information in almost all fields, be it in government, politics,

economics, society, culture, and even education. This is a strong reason for social media to

be one of the inseparable needs of today's society.

Based on data on internet and social media trends in 2021 in Indonesia researched

by We Are Social, it shows that the number of internet and social media users in Indonesia

has increased significantly in proportion to the increase in population. This proves that the

position of the internet and social media has shifted into an aspect that is inherent in the

daily activities of people in Indonesia (S. Raj, 2021). The following table presents data on

internet and social media trends for 2021 and 2022 in Indonesia.

No Information 2021 2022 Difference Differenc

e (%)
1. Population 268.2 Million 272.1 Million 3.28 Million 1%
 2. Internet user 150 Million 175.4 Million 25.4 Million 17%

3. Active Social 150 Million 160 Million 10 Million 1%

Media User

Based on the data above, it can be concluded that internet users in Indonesia account

for 64% of the total population, and this figure has increased from 56% in the previous

year. The phenomenon of increasing internet and social media users above is proof that the

internet medium has changed the pattern of human communication from time to time.

Services available on the internet lately tend to be very widely used by members of the

public to carry out communication activities, known as services in the form of social

networking applications (social network services). The social network service sites that

exist on the internet are indeed focused on efforts to build online communities from people

who want to share interesting experiences regarding the activities they carry out.

Domination of social media that is widely used, such as YouTube, which is indeed a

favorite among Indonesian people with number of users accessing YouTube, around 88%,

followed by social media Whatsapp with total access of 84%, Instagram with 79% and

followed by social media Facebook with a total 79% of this data is explained by one

website databoks.com (Shahbaznezhad et al., 2021). It can already be seen that social media

users through internet networks in Indonesia are quite large and their users also vary,

starting from 16 to 64 years of age. Many people use mobile social media with various

interests in it and various people who can reach them, this can have several impacts on the

public such as lots of hoax news and even less educational content.

REVIEW OF LITERATURE
 There is a grouping of data that has been divided into two objectives which refer to

clustering which can provide an understanding and clustering to be used. If you have a goal

to be able to understand, then the cluster that has been formed must be able to capture the

natural structure of a data (Soatto, 2020). Usually there is a process in clustering that has

the same goal and is an initial process which will then be followed by a summarization

process (mean, standard deviation), class labeling for each group which will be used as

training data related to classification, and so on (Kalam, 2020). Meanwhile, if there is a

goal to be used, then the main goal is to find the prototype cluster which is the most

representative value for a data and can provide an abstraction for each data object in a

cluster where the data is located. Segmentation is the process of grouping all heterogeneous

markets into a group or segment that has something in common in the desires and responses

to a strategic marketing program (Aditya et al., 2020). Customer segmentation is a group of

customers with the same needs and desires, in the segmentation process a grouping of

customers with the same values needs and characteristics occurs, as follows:

1. Demographic Segmentation

Statistics that can measure an observable aspect of the population, such as

height, age, and gender in terms of income, education and employment. This

demographic factor is the most common basis for classifying consumer groups.

2. Geographical

Geographical segmentation will be able to divide the market into a statistical

unit that can measure an observable aspect of the population, such as the presence of

height, age, and gender which concurrently in terms of income, education and
 employment. This demographic factor is the most common basis for classifying

consumer groups.

geographical areas such as countries, states, regions or neighborhoods. Companies

can choose to do business in one or more geographic countries and according to

their needs and desires to carry out marketing activities. This can be done by

dividing the market into various geographic units according to their needs and

wants, such as density and region.

3. Psychographic Grouping

Psychology by using demographic data that can determine a behavior and

also the preferences of groups in certain populations. Psychology can study

lifestyles, which are like having to go on vacation. Psychology is also included in

dynamic attribute segmentation, a way of looking at the market based on dynamic

characteristics that can reflect the characteristics of customers.

4. Behavioral Segmentation

In consumer market segmentation, that is something that has been based on a

segmentation of behavior. This has been done by grouping consumers into a market

segment based on behavioral or behavioral variables from the community that can

influence and also reflect knowledge and reactions related to products.

RESEARCH METHOD

This study has a population of 272,682.5 thousand Indonesian residents based on

statistical publications in 2022. Based on the sample calculation above, the number of

samples that will be used as respondents in this study is 399,9995, but will be rounded up to

400 respondent. Then, in the sampling technique that will be used, namely by using the
probability sampling method, where the sampling technique can provide an equal

opportunity for every member of the population that has been selected as a sample. In this

technique there are ways of selecting samples, namely by random random sampling,

systematic sampling and cluster sampling.

Based on this explanation, this study aims to identify user characteristics, product

categories and target segmentation from social media use using the K-Means Clustering

algorithm, which as a marketing plan strategy should be able to market a new product or

new feature. to attract the interest of the audience (Nadanyiova, 2020). K-Means Clustering

is one of the popular Machine Learning algorithms, especially Unsupervised Learning. In

this algorithm, you will be familiar with the value of k, which is the number of clusters to

be formed. K Means clustering is an unsupervised machine learning algorithm used for data

grouping and pattern recognition. The trick is to randomly select several initial data points

(k), then move them around until the most ideal grouping is found (Pan et al., 2020).

RESULT AND DISCUSSION

In the table below, the author presents the most chosen/most frequently used social

media by respondents arranged from the most to the least.

No Social Media Amount Persentase

1 Tik Tok 444 63%
2 Instagram 188 26%
3 Twitter 46 6%
4 Facebook 32 5%
Amount 710 100%
 Based on the table above, it can be seen that Tik Tok ranks first as the social media

most frequently used by respondents, namely 444 people with a percentage of 63%.

Instagram is in second place with 188 people or 26% and followed by Twitter with 46

people or 6%. The fourth position is occupied by Facebook with 32 people or 5%.

Cluster Analysis with K-Means

This study uses 3 variables to be able to segment related to the 4 biggest social

media platform customers in Indonesia using K-Meand Clustering. These three variables

can be divided into User Characteristics (X1), Content Categories (X2), and Updated News

Categories (X3). Prior to conducting the K-Means Cluster test, the research data will be

converted into a dummy variable to facilitate the process of dividing users into groups.

a. Cluster 1

Cluster 1 has 2% of the total population, and its user is a woman named

Jasmine, aged 36-45 years, who works as a private employee/employee. Married

status and domiciled in Central Java Last education Masters, with a monthly

income of IDR 6,000,000-IDR 10,000,000. Jasmine chose to use Tiktok social

media because it is more updated regarding current news, and has clear and

clear features that satisfy the audience to watch videos on Tik Tok. The selected

favorite content is about skincare and make up.

b. Cluster 2
Cluster 2 has 9% of the total population, and its users are a woman

named Karina, aged 37-48 who works as a Civil Servant. Married status and

domiciled in West Java Last education S1 with a monthly income of IDR

3,000,000-IDR 6,000,000. It's more enjoyable to use Instagram social media

because you can see the stories of the role models clearly and intensely, and

there are lots of choices for shopping for products from skincare, clothing to

even gadgets on Instagram social media. The most favorite Instagram account is

Nia Ramadhani's account because it is considered to be very influential by her

fans.

c. Cluster 3
Cluster 3 has 12% of the total population, and its user is a woman named

Rizfany, aged 15–24 years, who works as an employee or private employee.

Status: not married (single) and domiciled in DKI Jakarta. Graduated from high

school with monthly income of IDR 3,000,000-IDR 6,000,000. Choosing to use

Tik Tok social media more often because it is more useful related to fashion

trends and updates in the current era, and is available from various tips and

tricks. The most favorite content on Tik Tok is a day in my life for Tik Tok

celebrities or artists who appear on their FYP Tik Tok.

 d. Cluster 4
Cluster 4 has 13% of the total population, and its user is a woman named

Nindita, aged 25–34 years, who works as a private employee. Married status and

domiciled in DKI Jakarta. The last education is S1 with a monthly income of

more than IDR 10,000,000. Prefer to use social media Twitter because updates

on trends and news are more hype and faster on Twitter, and in this social media

the users are not as bothered as social media users on other platforms, Twitter

with its appearance makes the audience feel at home using this social media

every day. My favorite account on Twitter is Tanya RL because I always update

current news.

REFERENCE

Aditya Wilman, M., Amir Mahfuzh, M., Ervan Firdaus, M., Fithyan, M., Rifqi Muzakki, M., &
Sunan Gunung Djati Bandung, U. (2020). Analisis Aspek Pemasaran Segmentation,
Targeting, Brand Positioning Dan Bauran Pemasaran Pada Cocoes Indonesia.
 AKSELERASI: Jurnal Ilmiah Nasional, 2(1), 28–43.

Boediman, S. F., Hendriarto, P., Satmoko, N. D., Sulistiyani, S., & Sani, A. (2021). The
Relevance of Using Social Media Applications Strategies to Increase Marketing Potential of
Indonesian Maritime Tourism (Analytical Study of Tourism Journals and Online
Marketing). Budapest International Research and Critics Institute-Journal (BIRCI-
Journal), 4(4), 8791–8799.

Kalam, K., & Kalam, K. K. (2020). Market Segmentation, Targeting and Positioning Strategy
Adaptation for the Global Business of Vodafone Telecommunication Company.
International Journal of Research and Innovation in Social Science, IV(VI), 2454–6186.

Li, F., Larimo, J., & Leonidou, L. C. (2021). Social media marketing strategy: definition,
conceptualization, taxonomy, validation, and future agenda. Journal of the Academy of
Marketing Science, 49(1), 51–70. https://doi.org/10.1007/s11747-020-00733-3

Nadanyiova, M., & Das, S. (2020). Millennials as a Target Segment of Socially Responsible
Communication within the Business Strategy. Littera Scripta, 13(1).
https://doi.org/10.36708/littera_scripta2020/1/8

Pan, F., Shin, I., Rameau, F., Lee, S., & Kweon, I. S. (2020). Unsupervised Intra-Domain
Adaptation for Semantic Segmentation through Self-Supervision. Proceedings of the IEEE
Computer Society Conference on Computer Vision and Pattern Recognition, 3763–3772.
https://doi.org/10.1109/CVPR42600.2020.00382

S, S., & S. Raj, J. (2021). Analysis of Deep Learning Techniques for Early Detection of
Depression on Social Media Network - A Comparative Study. Journal of Trends in
Computer Science and Smart Technology, 3(1), 24–39.
https://doi.org/10.36548/jtcsst.2021.1.003

Shahbaznezhad, H., Dolan, R., & Rashidirad, M. (2021). The Role of Social Media Content
Format and Platform in Users’ Engagement Behavior. Journal of Interactive Marketing, 53,
47–65. https://doi.org/10.1016/j.intmar.2020.05.001

Yang, Y., & Soatto, S. (2020). FDA: Fourier domain adaptation for semantic segmentation.
Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern
Recognition, 4084–4094. https://doi.org/10.1109/CVPR42600.2020.00414