IIJDWM
Journal homepage: www.ifrsa.org
Data Mining: A Tool for the Enhancement of Banking
Sector
Shipra Kalra,Rachika Gupta
Lecturer, Chanderprabhu Jain College of Higher Studies and School of Law,
Sector A-8, Narela, Delhi
as economic indicators, competition, and customer
ABSTRACT
demographics [13]. Data mining can help these
Data mining is emerging as a very useful tool for organizations to better serve their customers, and
providing valuable information from large databases increase the effectiveness of the organization
and enabling managers and business executives to (Chopoorian et al., 2001). [8]
make hard core decisions in a much easier and Data mining is becoming strategically important for
effective manner. It is a process of analyzing the data banking sector as well. It analyzes the data from
from various perspectives and summarizing it into various perspectives and summarizes it into valuable
valuable information. This paper defines what data information. Data mining assists the banks to look for
mining is and how does it works. It then focuses on hidden pattern in a group and discover unknown
the use of data mining in banking industry on various relationships in the data. Today many banks are
areas like market segmentation, direct mail employing data mining for their proper functioning,
marketing, customer churn, fraud detection and risk the list includes: Bank of America, First USA Bank,
management. Headlands Mortgage Company, FCC National Bank,
Keywords: data mining, market segmentation, direct Federal Home Loan Mortgage Corporation, Wells
mail marketing, customer churn, fraud detection, risk Fargo Bank, Nations-Banc Services, Mellon Bank
management N.A., Advanta Mortgage Corporation, Chemical
Bank, Chevy Chase Bank, U.S. Bancorp, and USAA
1. INTRODUCTION
Federal Savings Bank. [11]
Almost every organization today is recognizing the This article begins with: what is data mining, how
importance of utilizing data mining in their data mining works. It then specifically focuses on
businesses. Data mining is emerging as a very useful what data mining can do in banking sector relating to
tool for providing valuable information from large Market segmentation, Customer churn, Fraud
databases and enabling managers and business detection, Risk management, Direct mail marketing,
executives to make hard core decisions in a much and Trend analysis.
easier and effective manner. Not only data mining
2. WHAT IS DATA MINING?
helps to extract useful knowledge from large amount
of data but it is also helping in declining various costs Data Mining is the process of extracting knowledge
like cost of computation power and also reduces time hidden from large volumes of raw data. According to
for decision making and knowledge discovery. Cabena et al. (1998) data mining is defined as the
Data mining is primarily used today by organizations process of extracting previously unknown valid and
with a strong consumer focus - retail, financial, actionable information from large databases and then
communication, and marketing organizations. It using the information to make crucial business
enables these organizations to determine relationships decisions [7]. Data mining software is one of a
among "internal" factors such as price, product number of analytical tools for analyzing data. It
positioning, or staff skills, and "external" factors such allows users to analyze data from many different
IFRSA International Journal of Data Warehousing & Mining |Vol1|issue 2|Nov 2011 204
Shipra Kalra, Rachika Gupta|Data Mining: A Tool for the enhancement of Banking Sector
dimensions or angles, categorize it, and summarize indebtedness periods, generating more than $25
the relationships identified. Technically, data mining million in additional interest earned.
is the process of finding correlations or patterns
3. HOW DATA MINING WORKS?
among dozens of fields in large relational databases.
Data mining tools predict behaviors and future 1. Understanding the Business: The objectives and
trends, allowing businesses to make proactive, scope of the business are determined and the job to
knowledge-driven decisions. Data mining tools can be done is converted into a data mining problem.
answer business questions that traditionally were too 2. Understanding the data: The data is collected
time consuming to resolve. They search databases for which includes: operational or transactional data, non
hidden patterns, finding predictive information that operational data, and metadata. The patterns,
experts may miss because it lies outside their associations, or relationships among all this data are
expectations. found out.
Data mining technology can generate new business 3. Extraction of data: Extract, transform, and load
opportunities by: transaction data onto the data warehouse system.
Automated prediction of trends and behaviors: Data 4. Storage: Store and manage the data in a
mining automates the process of finding predictive multidimensional database system.
information in a large database. Questions that 5. Access: Provide data access to business analysts
traditionally required extensive hands-on analysis can and information technology professionals.
now be directly answered from the data. A typical 6. Modeling: Various modeling techniques like
example of a predictive problem is targeted decision trees, clustering, visualization, predictive
marketing. Data mining uses data on past modeling etc are applied to make knowledgeable
promotional mailings to identify the targets most decisions.
likely to maximize return on investment in future 7. Presentation of Data: Present the data in a useful
mailings. Other predictive problems include format, such as a graph or table.
forecasting bankruptcy and other forms of default,
4. BANKING SECTOR
and identifying segments of a population likely to
respond similarly to given events [4]. As banking competition becomes more and more
Automated discovery of previously unknown global and intense, banks have to fight more
patterns: Data mining tools sweep through databases creatively and proactively to gain or even maintain
and identify previously hidden patterns. An example market shares. The banks of the future will use one
of pattern discovery is the analysis of retail sales data asset which is knowledge for their existence and
to identify seemingly unrelated products that are excellence. That is why the banking industry is
often purchased together. Other pattern discovery recognizing the importance of information and has
problems include detecting fraudulent credit card started using information technology not only to
transactions and identifying anomalous data that improve the quality of service, but also to gain a
could represent data entry keying errors. [4] competitive advantage. The enormous amounts of
For example: A bank searching for new ways to data that banks have been collecting over the years
increase revenues from its credit card operations can be effectively mined so that the bank executives
tested a non intuitive possibility: Would credit card can predict with increase accuracy, how customers
usage and interest earn increase significantly if the will react to adjustments in interest rates, which
bank halved its minimum required payment? With customers will be likely to accept new product offers,
hundreds of gigabytes of data representing two years which customers will be at higher risks for defaulting
of average credit card balances, payment amounts, on a loan and how to make each customer
payment timeliness, credit limit usage, and other key relationship more profitable (Fabris 1998) [9].
parameters, the bank used a powerful data mining The following sections will describe how data mining
system to model the impact of the proposed policy can be effectively used in various fields of banking
change on specific customer categories, such as like:
customers consistently near or at their credit limits
4.1. Market segmentation
who make timely minimum or small payments. The
bank discovered that cutting minimum payment Identify the common characteristics of customers
requirements for small, targeted customer categories who buy the same products from your company.
could increase average balances and extend Banks deal with various types of customers e.g.,
individuals, group of people, corporate entities, etc.
IFRSA International Journal of Data Warehousing & Mining |Vol1|issue 2|Nov 2011 205
Shipra Kalra, Rachika Gupta|Data Mining: A Tool for the enhancement of Banking Sector
who have their likes and dislikes. No bank can afford With the help of data mining technique, banks are
to assess the need of each and every individual able to analyze which products and service are
customer separately. It is nearly impossible for banks availed by most of the customers in cross selling and
to market all these categories of customers on a one- which type of consumers prefer to purchase cross
to-one basis, particularly if they simply rely on selling products and so on. For example: During
predictable socio-economic data like age and income 2008, there was a 37% increase in banking direct
as the base for dividing customers into segments. To mail pieces sent to manage current client
overcome this problem, banks marketing department relationships, including informational and loyalty
use data mining for market segmentation strategy, mailings, renewal notices and upgrade offers. E-mail
which recognizes the wisdom of specializing to suit cross-selling also increased. Cross-sell e-mails
the need of a segment of the market rather than trying tracked through Mintel Comperemedia's e-mail panel
to address the requirements of each and every rose from a 2% share of banking e-mail in 2007 to a
customer separately. Market segmentation divides the 5% share in 2008 [16].
whole market into groups of customers who have the
4.3. Customer Churn
requirement of similar kinds of products and services.
Each segment of the market may demand different High cost of customer acquisition and customer
products and require different marketing mix to education requires companies to make large upfront
address the demand. The bank should, therefore, investments on customers. However, due to easy
develop the profile of different market segments. access to information and a wide range of offerings,
Then the targeted market segments should be selected it is easier than ever before for customers to switch
based on their attractiveness. Once the bank has between service providers. This applies to all
identified the market segments that it might address, industry verticals such as banking, telecom,
the next steps will be positioning of the product into insurance, etc. customer churn, which is defined as
the targeted market segment. For example: The first the propensity of customers to cease doing business
major step in the direction of marketing was initiated with a company in a given time period, has become a
by The State Bank of India in 1972, when it significant problem and is one of the prime
recognized itself on the basis of major market challenges financial institutions worldwide are
segments, dividing the customers on the basis of learning to face [6].
activity and carved out four major market segments. In banking domain, we define a churn customer as
They are commercial and institutional segment, small one who closes all his/her accounts and stops doing
industries and small business segment, agriculture business with the bank. There are many reasons for a
segment and personal and services banking segment customer to close the account(s). For example, a
[5]. person creates an account for a specific purpose and
closes it immediately after the purpose is solved. Or a
4.2. Direct mail marketing
person is relocated and has to move to another place
Identify which prospects should be included in a and hence closes all the accounts. Or a customer may
mailing list to obtain the highest response rate. In the stop transacting with the bank just because of the
banking sector, direct mail marketing has unavailability of banks ATMs in important places
traditionally been very popular. The use of and hence close his/her accounts [6]. Banks can use
transpromo mail where offers are included in data modeling techniques like fuzzy methods,
transaction mail has remained consistent through predictive modeling can identify the customers that
the economic recession. With many banks struggling are likely to churn in the near future and then the
to retain their customers, transpromo direct mail has bank executives can provide exciting offers to these
played a critical part in their marketing efforts [15]. customers which they cannot refuse for example, a
Banks can use data mining to analyze customer mortgage customer may tell the lending bank about
databases and develop profiles of individual customer an existing auto loan from another bank. An agent of
preferences for products and services. They can offer the bank can add this information to the customers
those products and services that the customer really profile, and present back a pre-approved refinance of
wants by directly mailing and cross selling i.e. the auto loan. This will save the customer money by
banks makes an attractive offer to its customer by consolidating the existing mortgage and auto loan
asking them to buy additional product or service. For with one bank and also prevent the customer from
example home loan with insurance facility. searching service offerings from other banks.
IFRSA International Journal of Data Warehousing & Mining |Vol1|issue 2|Nov 2011 206
Shipra Kalra, Rachika Gupta|Data Mining: A Tool for the enhancement of Banking Sector
4.4. Risk management BANK to detect suspicious credit card transactions.
According to Kuykendall Flacon is used by 9 off the
Banks manage risks relating to offering new
top ten credit card issuing banks, where it examines
customers credit cards, extending existing customers
the transactions of 80% of cards held in the U.S.
lines of credit, approving loans, amount of loan,
Mellon Bank also uses data mining for fraud
lending rate, repayment period, loan defaults etc. For
detection.[11]
example: Bank executives by using data mining
technique can reduce the risk of issuing credit cards 5. CONCLUSION
by determining those customers who are likely to
This article described that data mining can be a very
default on their accounts.
powerful and helpful tool to extract important and
Banks also have the problem of predicting the credit-
useful information for banking sector from the
worthiness of new clients on the basis of historic data
historical as well as from the current data. Data
of past clients. The creditworthiness also influences
mining can be used in various fields of banking like
the interest rate of a credit. A bank has data about
Market segmentation by which banks can segment
clients to whom it gave credits in the past. The client
their customers into different groups, direct mail
data contain personal data, data describing the
marketing can help the banks to improve their
financial status and the financial behavior before and
marketing strategy and to increase their business,
at the time the client was given the credit. The clients
customer churn to increase the rate of retention of the
are divided into four classes. The first class contains
customers, risk management to reduce the various
all those clients who paid back the credit without any
risks like creditworthiness and fraud detection to
problems; the second class those who paid back with
reduce the number of fraudulent. Data mining has
little problems; the third contains those who should
wide application domain in almost every industry
only get a credit after detailed checks because
where bulky data is generated and that is why it is
substantial problems of payback occurred in the past;
consider as one of the most important and promising
and the forth class consists of all those who did not
developments in Information Technology.
pay back at all. Using this data table a prediction
model is created in order to predict the probability for REFERENCES
each class for new clients. By this way the
[1] Rene T. Domingo; Applying data mining to
combinations of attributes which are responsible for
banking.
clients to have a high probability of not paying back
[2] Vivek Bhambri,Dept. of Computer Sciences, Desh
will be identified by the prediction model too [18]. Bhagat Institute of Management and Computer
4.5. Fraud Detection Sciences, Mandi Gobindgarh, Punjab, India;
Application of Data Mining in Banking Sector.
In banking, fraud can involve using stolen credit [3] Amir M Hormazi, Information Systems
cards, forging checks, misleading accounting Management; spring 2004; 21, 2; ABI/INFORM
practices, etc. Data mining can help to detect Global; Data Mining a competitive weapon: for
fraudulent actions by building a model using a banking and retail industries.
fraudulent behavior that has been done in the past and [4] Doug Alexander Dea; Data Mining.
then use data mining to identify similar behavior. [5] Beehive digital concepts Cochin for Mahatma
Through the use of sophisticated data mining tools, Gandhi University Kottayam; Marketing strategies
millions of transactions can be searched to spot of the banking industry.
patterns and detect fraudulent transactions. Various [6] M Purna Chandar, Arijit Laha, and P Radha
techniques like decision trees (e.g. CHAID), machine Krishna; Hyderabad; Modeling churn behavior of
learning, association rules, cluster analysis, neural bank customers using predictive data mining
networks and predictive models can be generated to techniques.
estimate things such as probability of fraudulent [7] Cabena, P.Hadijinian, P.Stadler, R.Verhees and
behavior or the dollar amount of fraud. Zanasi; Prentice Hall, New Jersey; Discovering
Data Mining: From Concept to Implementation.
In banking, the most widespread tool used to detect [8] Chopoorian, J.A., Witherell, R.Khalil, O.E.M and
fraud is HNCs Falcon product .HNC monitors more Ahmed; SAM Advanced Management Journal;
than 160 million payment-card accounts to detect Mind your business by mining your data.
credit card frauds. They also claim a healthy return [9] Fabris, P.1998; Advance Navigation.
on investment. HNC was built using a neural network [10] Kuykendall, L. September 1999; The Data mining
shell and is used by many banks like LLOYDS tool box: Credit Card Management.
IFRSA International Journal of Data Warehousing & Mining |Vol1|issue 2|Nov 2011 207
Shipra Kalra, Rachika Gupta|Data Mining: A Tool for the enhancement of Banking Sector
[11] Ch08.fm Page 191 Monday, September 6, 1999 [15] http://www.overnightprints.com/news/direct-mail-
10:11 AM; Industry Applications of Data Mining. usage-in-banking-and-investment-industries-
[12] David H. Pyle; University of California; Berkley; grows-956
Bank Risk Management: Theory. [16] http://www.dmnews.com/banks-increase-direct-
[13] Data Mining: What is Data Mining? marketing-to-customers/article/129580/
http://www.anderson.ucla.edu [17] http://www.statsoft.com/textbook/fraud-detection/
[14] http://www.ekantipur.com/the-kathmandu- [18] http://www.easydatamining.com
post/2010/03/01/Oped/Plastic-pollution/5702/
IFRSA International Journal of Data Warehousing & Mining |Vol1|issue 2|Nov 2011 208
