International Conference on Innovative Mechanisms for Industry Applications

(ICIMIA 2017)

An Efficiency Analysis on the TPA Clustering

Methods for Intelligent Customer Segmentation
Dr. (Mrs) Ananthi Sheshasaayee1 L.Logeshwari2
1
Research Supervisor, 2Research Scholar
PG & Research Department of Computer Science, Quaid -E- Millath Govt College for Women (Auto)
Chennai-600 002, India.
1
[email protected], [email protected]

Abstract demographic, RFM, and LTV are used to determine CRM

segmentation process.
Now a day’s commercial marketing growth is improved by
customer segmentation model. Literatures use the data
mining technology to review the customer segmentation
and sound effectives. Stages of CRM have been used in
most of the cases. Based on RFM, demographic and LTV
data the paper is prepared using the data mining tools for
the new customer segmentation. Customer segmentation
has been done by RFM and LTV method. By using K-
means clustering and demographic data, the customers are
clustered into different categories. LTV profile of the
customer is finally ready. This method has been followed
by various other industries too.
Keywords: customer relationship management, segmentation,
data mining, clusters. Fig. framework for customer segmentation
Introduction
Customer segmentation is done based on group of An innovative approach for combing to all variable like
individuals who are relevant to marketing based on their age, data mining clustering retention of customers and
gender, interests and spending habits. Customer Relationship development are done using CRM, RFM, and LTV.
Management (CRM) has become a leading in business The proceeds of this paper are done as follows.
environment. To manage the business opportunities with Section two: previous studies are taken into account for
customers managerial efforts are taken (Kim, Suh, & Hwang, customer segmentation and this explains the reasons for the
2003). Some of the benefits of customers are background of that particular customer.
(1) customer’s continuous retention Section three is about clustering model for customer
(2) loyalty of the customer segmentation. Subsequently the result of the model in banking
(3) higher profit giving customer industry will be submitted.
(4) new ideas creating customer Finally, Section four: it will end with the paper on general
(5) Using of more product and service customer are discussions and an agenda.
done. II. LITERATURE REVIEW
CRM followed these four dimensions. Customer segmentation and market segmentation are
(1)To find customer identity, (2) Customers charm, said to be same in literature but there are many differences in
clustering and data availability using CRM, new customer are
(3) To retaining customers, (4) Customers growth.
acquire new customer by using socio demographic data and
Customer identification is based on marketing transactional data. “The method can tell us that customer
researchers and it has a great contribution to the other steps of cultivation and retention are more important than customer
CRM [6].Loyalty program and profitable segments are allotted acquisition. Lack of information on new customers makes it
to Customers to attract them. The highest profit giving difficult to select target customers and this will cause
customers are retained and UN profitable customers are taking inefficient marketing efforts” [8]. Chai & Chan [9] customer
into account and the amount of resources given to them is segmentation both methodology and application oriented.
Most methods-driven studies changes some data clustering
reduced. Invaluable customers are a burden to the company
techniques such as SOM, or two or more combinational data’s
and so not a small amount is spent on them. Input variables like are used in data mining techniques and more accurate
978-1-5090-5960-7/17/$31.00 ©2017 IEEE 784
 International Conference on Innovative Mechanisms for Industry Applications
(ICIMIA 2017)
achievements in clustering segmentation. Many specific
segmentation problems are solved with application- oriented
researchers. (Such as [10], [6], [11], and [12]). [9]. To create
new variables based on clustering technique and use different
sequential clustering steps. (Such as [8], [7], [13], [14], [9],
[15], [16], [2], [17], [18], and [5]).
LTV place a significant role in customer
segmentation. Based on previous profit, potential benefit and
defection probability of customers the data are segmented. The
case study of a wireless telecommunication suggests a value of
customers. Then, at last strategies of customer segments are
explained through a case study on a wireless
telecommunication company.
The next important customer segmentation method is III. RESEARCH FRAMEWORK
RFM. The important customer segmentation is RFM, as it is This research framework consists of four stages of
based on customer’s repayment behavior and recency, customer segmentation by CRM. Some banking industry
frequency monitory and self-organizing map is prepared. implemented this type of framework. The bank customers and
By using an Apriori algorithm association rules the develop marking strategies were categorized by segmentation.
groups of customers were profiled by customer’s attributes. This research approach is clustered into two phases.
Two different data sets like RFM, CHAID and logistic
regression are investigated as analytical methods. Value of
RFM and k-means algorithms are utilized by researches to
extract the new meaning of the customers.
In example [9] describes the novel approach that
combines customer segmentation and customer target. RFM
and LTV models are identified based on customer behavioral
method. Different variables are used by many authors based
on clustering technique. The customer satisfaction survey is an
easy and practical approach. Many authors execute the multi
agent based system and survey based profitable system as a
tool to execute the customer satisfaction survey. The business
intelligent tools are data envelopment analysis (DEA), self-
organizing maps (SOM) are the tools which are used for
profitable customer segmentation. Customers purchasing
behavior method is done by anticipation model [14].
Clustering analysis and association rule analysis are
referred to know the past purchasing behavior of loyal and
potential customers. Based on the data mining and new
customer segmentation method the customer’s framework is
done. The new customer segmentations contain two types.
First methods, Customers are clustered by survival characters Fig. Research framework (TPA)
based on k-means clustering arithmetic method. Second
method is survival analyzing method based on cluster’s First to perform a two-phase method to segment
survival and customer churn trend.[7][8][9][10] bank customer data’s are very useful to find RFM and
In [15] online game customers are investigated by demographic variables. RFM is used in the first stage of the
data mining method and experimental marketing method. It input values of K-means clustering. The demographic data,
helps us to understand the customer’s purchasing behavior. based on the age, occupation and education is used to cluster
The manufactures then accordingly alters the market strategies stages of all data’s. These three are simultaneously used in
to increase the purchasing level of the customers. selection procedures of all data’s using SOM.
Earlier the segmentation methods were based on with
Authors have focused the segmentation procedure in segment and inter-segment heterogeneity and the markers
from the technical point of view example is the cross national were more interested only in making the net value of the
market segmentation method. The first phase is the statistical important customers, but now to use neural network to
method under which the difference between the clustering is calculate the potential value of all the customers by their LTV
done and the second phase of data mining method is to and finally going through their profile.
develop the clustering system within each nation, for this
support vector clustering is used. [15][16][17].
For online shopping market marketing segmentation
is used and novel clustering algorithm based on genetic
algorithms is used.
Table I categorizes segmentation models proposed by
different authors according to their input variables.
978-1-5090-5960-7/17/$31.00 ©2017 IEEE 785
 International Conference on Innovative Mechanisms for Industry Applications
(ICIMIA 2017)
ti customer index for service period i
IV. FORMULA FOR ANALYSING DATAS Ni period of service the customer rate of interest i
d interest rate
A. Reliable Data and measurement E (i) total service period expected by the customer i
For example in case study of an Iranian bank to πp(ti) contribution of past profit of the customer ti
make decisions for satisfying the customer, the bank cross πf (ti) contribution of future profit of the customer ti and
sold the opportunities and loyalty programs. Some important B (ti) the customers potential benefit i at period ti
records were taken from the customer’s data and were To calculate the current value and potential value of
discarded. the customer based on LTV method. So the balance sheets of
After an SOM clustering the demographic variables each customer for the last six months are to be calculated
on the clustering procedures were ranked. According to the according to their current value. Potential values are predicted
transactions of the customers RFM values were calculated. by many authors based on Eq.2
The age level, occupational level and education level
and education level were taken into account. Last 6 months ---------- (2)
transactions were taken as the data base. Thus a reliable data
was found. Customer Loyalty ¼ 1 _ Churn rate
B. Customer clustering using RFM and demographic
Variables Previously LTV models do not link any campaign
Customer segmentation was utilized by two-phase clustering. activities. Most of the researchers to correlate with customer
Customers are clustered based on their RFM values. As the value and campaign activities, which are main problem. Many
results, the three clusters were detected consisting of “190, authors do not consider LTV model. Calculating a new model
111, and 190 Customers”. Secondly, each cluster was customer value based on a serial number of campaigned
internally clustered based on three demographic variables on projects are already done, as follows
education, occupation and age. Therefore nine clusters were
the results consisting of “104, 52, 34, 118, 23, 49, 32, 42, and
37 customers”.

“Where the Probij is the probability of marketing

strategy j is for segmented customer i, Mij is the contribution
factor of increasing of increasing monetary value of
segmented customer i for segmented customer i and N is the
customer loyalty expected in the year.”

--------- (3)
Lij contributes factors of increasing loyalty of the customer
While j is the adapting strategy, i customer strategy and M is
Number of marketing strategy of the customers.

The new technology model to add two contribution

factor variables are The first variable are the probability of
customer i adapting Strategy j, and the other factors are
contribution value when customer i uses strategy j. These
variables linked Campaign programs and customers value.
Similarly to evaluate potential values are
interviewing with managers, to found the kind of new services
that could be proposed to their customers in the future and the
return on each services. For example, a new service in ATM,
Fig. information flow of value based customer segmentation. the bank managers only recommend to transferring money and
C. Clusters using LTV analysis bill payments of their accounts. Then they predicted an
In [8] based on new LTV model considered churn rate of a average profit which would be gained from these new
customer is done. This model is shown in Eq. 1. services. Using neutral network the occurrence profitability of
new services was predicted. The neural network model
classification function is done by prediction model based on
customer’s behavioral pattern regarding to new recommended
service.

978-1-5090-5960-7/17/$31.00 ©2017 IEEE 786

 International Conference on Innovative Mechanisms for Industry Applications
(ICIMIA 2017)
and finding opportunities the customer relationship
management further do various researches.
The aim of CRM is to make a data based on the two
phase clustering model, life charts and finally they use neural
network to enhance the performance of the customer and
compare and reach to the best selection of all.

REFERENCES
[1] Jutla, D., Craig, J., & Bodorik, P., “Enabling and measuring electronic
Customer relationship management readiness”, proceedings of the 34th
annual Hawaii international conference on system sciences organizational
Systems and technologies track (pp. 1–10), 2001.
[2] Stone, M., Woodcock, N., & Wilson, M., “Managing the change from
Marketing planning to customer relationship management”, Long Range
Planning, 29, 675–683, 2006.
[3] Swift, R. S.” Accelerating customer relationships: Using CRM and
Fig. customer life time value computation Relationship technologies”, Upper saddle river N.J.: Prentice Hall PTR,
2001.
[4] Parvatiyar, A., & Sheth, J. N., “Customer relationship management:
In new service usage the customers predict “yes” or
Emerging practice, process, and discipline”, Journal of Economic & Social
“no” in the confidence of model method. This confidence is Research, Vol.3, pp. 1–34, 2001.
used to calculate the Prob ij. Customer’s potential value [5] Kim, Yong Seog, Street, W. Nick, Russell, Gary J., & Menczer,
prediction is used for two essential items, such as Probij and Filippo, “Customer Targeting: A Neural Network Approach Guided by
Genetic Algorithms”, Management Science, 51(2), 264–276, 2005.
Profitij. Prediction LTV for each customer is used to calculate [6] Jonker, Jedid-Jah, Piersma, Nanda, & Poel, Dirk Van den, “Joint
means value based on prediction rule of the customer. As a optimization of customer segmentation and marketing policy to maximize
result of this nine different cluster of two phase segmentations long-term profitability”, Expert Systems with Applications, 27, 159–168,
are found. 2004.
[7] Su-Yeon Kim, Tae-Soo Jung, Eui-Ho Suh, Hyun-Seok Hwang,
“Customer segmentation and strategy development based on customer
lifetime value: A case study” Expert systems with applications, Vol. 31, pp.
101-107, 2006.
[8] Hyunseok Hwang, Taesoo Jung, Euiho Suh , “An LTV model and
customer segmentation based on customer value: a case study on the
wireless telecommunication industry” Expert systems with applications,
Vol. 26, pp. 181-188, 2004.
[9] Chu Chai Henry Chan,“Intelligent value-based customer segmentation
method for campaign management: A case study of automobile retailer”,
Expert systems with applications, Vol. 34, pp. 2754-2762, 2008.
[10] Sang Chul Lee, Yung Ho Suh, Jae Kyeong Kim, Jun Lee, “A
Cross-national market segmentation of online game industry using SOM”,
Expert systems with applications, Vol. 27, pp. 599-570, 2004.
[11] Jih-Jeng Huang, Gwo-Hshiung Tzeng, Chorng-Shyong Ong,
“Marketing segmentation using support vector clustering”, Expert systems
Applications of management
With applications, Vol. 32, pp. 313–317, 2007.
The cluster can be found by going through the profile [12] Kyoung-jae Kim, Hyunchul Ahn, “A recommender system using GA
of the customers. This profile is shown in Table II. These K-means clustering in an online shopping market”, Expert systems with
ranks show the clustering of all the ranks. For Applications, Vol. 34, pp. 1200–1209, 2008.
Example, cluster 1 of RFM is placed in the sixth position it [13] Nan-Chen Hsieh, “An integrated data mining and behavioral scoring
Model for analyzing bank customers”, Expert systems with applications,
means that clusters are important to this cluster RFM. Vol. 27, pp. 623-633, 2004.
Occupation level of this cluster is dominated over the other [14] Horng-Jinh Chang, Lun-Ping Hung,*, Chia-Ling Ho, “An anticipation
clusters and it’s the best. model of potential customers’ purchasing behavior based on clustering
So regarding education this cluster is consider to be analysis and association rules analysis” Expert systems with applications,
Vol. 32, pp. 753-764, 2007.
the best (rank 2). It is also the largest cluster when compared [15] Jyh-Jian Sheu, Yan-Hua Su, Ko-Tsung Chu, “Segmenting online game
with others because of the number of customers it has. (See customers – The perspective of experiential marketing”, Expert systems
table II) cluster profile is based on the table which indicates with applications, Vol. 36, pp. 8487–8495, 2009.
that marketers would be able to make decisions better to in [16] John A. McCarty, Manoj Hastak, “Segmentation approaches in
datamining:
improving their marketing strategies and show profit for their A comparison of RFM, CHAID, and logistic regression”, Journal
organization. of Business Research, Vol. 60, pp. 656–662, 2007.
[17] Jang Hee Lee, Sang Chan Park “Intelligent profitable customers
segmentation system based on business intelligence tools”, Expert systems
VI. CONCLUSION
with applications, Vol. 29, pp. 1[2]–152, 2005.
The most important target for marketing is to identify [18] Ching-Hsue Cheng, You-Shyang Chen, “Classifying the segmentation
the largest buyer and analyze their portfolios by segmentation of customer value via RFM model and RS theory”, Expert systems with
and marketing strategies. In this clustering the exiting applications, Vol. 36, pp. 4176–4184, 2009.
customers are divided into nine groups, based on the [19] Chihli Hung, Chih-Fong Tsai, “Market segmentation based on
Hierarchical self-organizing map for markets of multimedia on demand”,
characteristics of their transactions. Marking strategies are Expert systems with applications, Vol. 34, pp. 780-787, 2008.
analyzed in each group keeping in mind the profiles of the [20] Wedel, M., & Kamakura, W. A., “Market segmentation: Conceptual
customers. In order to improve the customer loyalty, revenue And methodological foundations” (2nd Ed.). Dordrecht: Kluwer, 2000.
[21] Irene Gil-Saura and Maria-Eugenia Ruiz-Molina, “Customer
segmentation based on commitment and ICT use”, Industrial Management
978-1-5090-5960-7/17/$31.00 ©2017 IEEE 787
 International Conference on Innovative Mechanisms for Industry Applications
(ICIMIA 2017)
& Data Systems, Vol. 109 No. 2 pp. 206-223, 2009.
[22]Duboff, R. S. (1992). Marketing to maximize profitability. The Journal of
Business Strategy, 13(6), 10–13.
[23]Dwyer, F. R. (1997). Customer lifetime valuation to support marketing
decision
making. Journal of Interactive Marketing, 11(4), 6–13.
[24]Gloy, B. A., Akridge, J. T., & Preckel, P. V. (1997). Customer lifetime
value:
An application in the rural petroleum market. Agribusiness, 13(3),
335–347.
[24] Pham, D.T. and Afify, A.A. (2006) “Clustering techniques and their
applications in engineering”.
Proceedings of the Institution of Mechanical Engineers, Part C: Journal of
Mechanical Engineering
Science,
[25] A.K. Jain, M.N. Murty, and P. J. Flynn,(1999) “Data clustering: a
review”, ACM Computing Surveys
(CSUR), Vol.31, Issue 3,, 1999.
[26] Grabmeier, J. and Rudolph, A. (2002) “Techniques of cluster algorithms
in data mining” Data Mining
and Knowledge Discovery, 6, 303-360.
[27] Han, J. and Kamber, M. (2001) “Data Mining: Concepts and
Techniques”, (Academic Press, San
Diego, California, USA).
[28] Jiyuan An , Jeffrey Xu Yu , Chotirat Ann Ratanamahatana, Yi-Ping
Phoebe Chen,(2007) “A
Dimensionality reduction algorithm and its application for interactive
visualization”, Journal of Visual
Languages and Computing, v.18 n.1,, February p.48-70.
[29] Nargess Memarsadeghi, Dianne P. O'Leary,(2003) “Classified
Information: The Data Clustering
Problem”, Computing in Science and Engineering, v.5 n.5, September p.54-
60,

978-1-5090-5960-7/17/$31.00 ©2017 IEEE 788