You are on page 1of 4

Banking Intelligence: Application of Data

Warehouse in Bank Operations


Zhonglin Lin, Minghua Zhu, Wenjun Yin and Jin Dong
IBM China Research Lab
Daimond Building, ZGC Software Park
Haidian District, Beijing 100193, PRC
Email: {linzl.zhumingh.yinwenj.dongjin}@cn.ibm.com

Abstract-Along with the development of information technol- the bank intelligence in real world bank business. Our solution
ogy, business intelligence plays an important role in the bank focus on the function of customer analysis and branch location
operation process. Bank Intelligence is a method of storing and analysis. The business is implemented by taking advantage
presenting key bank business data so that anyone in the bank can
quickly and easily ask questions of accurate and timely data. In of OLAP technology for analyzing data maintained in data-
a bank network, hundreds of millions of customer data records warehouses in our solution [2]. A huge amount of information
are generated daily. The growing competition and increased speed stored inside banking information systems about customers
of business changes has dramatically shown the need for bank and their transactions, but the banks can rarely exploit its
intelligence. In this paper, we propose an data warehouse based full potential. The more the data, the more difficult it is to
bank intelligence framework and implementation. Specifically,
we introduce the application of data warehouse in the analysis collect and transform them into useful knowledge. Our systems
of bank customer data and bank branch data. This paper tackle this problem in a data warehouse framework, the main
also summarizes experiences and results of applying this this advantage of our solution can be summarized as follows:
solution into real-world applications. Our results has revealed • Clear architecture. The solution employs the prevalent
the simplicity and power of OLAP-based solutions to scalable
bank intelligence. data warehouse concept. Data are collected into the data
Index Terms-banking, data warehouse, OLAP warehouse first, and then fed into our specially designed
data mart. The reason to use warehouse is for better
I. INTRODUCTION incorporation of our solution. Many banks have their own
data warehouse. Above those, various data marts can be
If banks are to succeed in today's markets, it is absolutely built for various purposes. Our solution aims to adapting
essential that their central key processes be designed efficiently the existed warehouse and can be quickly deployed using
and effectively. Bank intelligence is emerged to satisfy this this architecture.
requirement. Bank intelligence refers to the ability of gath- • Subject oriented. Based on our knowledge of banking
ering, managing, and analyzing of large amounts of data on industry, especially the domestic market, we summarize
a banks customers, products, services, operations, suppliers, an overall subject library. Those subjects are organized
and partners and all the transactions in between. Examples as main topic, sub-topic, etc. The bank, as their need,
of bank intelligence applications include the customer data can select any subject that interested. The subject ori-
analysis, branch location analysis, etc [1]. As a more and ented manner makes our solution unambiguous and cost
more important business activity manner, bank intelligence effective.
can provide personality services for customers and radically • Timely response. This benefit inherits from the OLAP
improve the service quality of bank. technology. Other than the off-line analysis and report,
Customer data analysis and branch location analysis are two our solution can trace the change happening and provide
key steps in bank intelligence. Customers should be at the valuable information by timely response.
heart of any business. These systems collect large volumes of
data about customers which contain valuable information that II. BANKING INTELLIGENCE AND OLAP
can allow a business to improve its customer relationships and Banking executives are focused on results. They need to
services. Typically, customer analytic focus on reporting what know the bank's customers in great depth so as to sell the
has happened and predicting the possible future behaviors of right product to the right customer for the right reasons. They
customers. The location of branches has an important influence must also adapt to pressures from globalization, integration,
on the performance of the bank business. Bank intelligence growing competition, product and market innovations. What's
should conduct an evaluation of the location of different more, they must also manage risk and comply with changing
branches in order to provide an insight into different factors requirements. All these challenges leads to the introduction of
that affect the performance. Thus analysis of location data is banking intelligence, which refers to technologies, applications
an important subtask of bank intelligence. and practices for the collection, integration, analysis, and
In this paper, we mainly introduce a solution which apply presentation of business information and also sometimes to

978-1-4244-2013-1/08/$25.00 ©2008 IEEE 143

Authorized licensed use limited to: GOVERNMENT COLLEGE OF TECHNOLOGY. Downloaded on February 27, 2009 at 09:44 from IEEE Xplore. Restrictions apply.
the information itself. Bank intelligence can provide historical, B. Design and Implementation of Data Warehouse
current, and predictive views of banking operations, most often Our system is built on the foundation of data warehouse
using data that has been gathered into a data warehouse or a techniques. The design of data data warehouse is the key factor
data mart. to the success of our system. Figure 1 shows the architecture
In our system, banking intelligence is mainly composed of the data warehouse. We can see that our solution, which can
of customer intelligence and location intelligence. Customer be implemented as a data mart, is built above the general data
intelligence can be defined as the act of capturing, managing, ware house. As the figure shows, our solution is only one of
analyzing and applying customer information in the bank the solutions which can be applied on the warehouse and the
for profitable growth. Infonnation can be obtained about architecture provide sufficient security and extension ability.
customers' existing and future needs, how they reach deci-
sions' about their behavior as well as about the competition,
conditions in the industry, and general trends. To properly
manage the relationship with the customer the bank needs to
collect the right information about its customers and organize
that information for proper analysis and action. Location
Intelligence is the capacity to organize and understand com-
plex phenomena through the use of geographic relationships
inherent in all information. By combining geographic- and
location-related data with other business data, organizations
can gain critical insights, make better decisions and optimize
important processes and applications.
OLAP(On-Line Analytical Processing) tools is now playing
a key role in the design and implementation of banking
intelligence. OLAP is an approach to quickly provide answers
to analytical queries that are multi-dimensional in nature. In
most cases, data mining techniques are used in partnership
with OLAP as an assisting tool. OLAP has achieved its success
in banking intelligence due to: Fig. 1. The data warehouse architecture.
• Fast Speed :means that the system is targeted to deliver
most responses to users in a totally acceptable time with We notice that the ETL (Extraction, Transformation, and
the necessary analysis. This speed is not easy to achieve Loading) plays an important role in the architecture. There
with large amounts of data, particularly if on-the-fly and exist various ETL tools which make the process automated and
ad hoc calculations are required. fast. As a result, we employ an automation tool to complete
• Analysis: means that the system can cope with any the task.
business logic and statistical analysis that is relevant for
C. Subject Template
the application and the user. It is certainly necessary to
allow the user to define new ad hoc calculations, and Based on the experience in banking industry, an analysis
allow the data be reported in any required forms to the subject template is developed. The knowledge is grouped
user. by topic and sub-topic. For each topic, a multidimensional
• Multi-dimensional: means that the system must provide analysis is designed. The key to these analysis is the set
a multidimensional conceptual view of the data. It is the of measures and dimensions. Figure 2 illustrates a part of
key factor and the most important requirement of OLAP. the subject template and Figure 3 gives an example of the
"measure-dimension" view of the star shape OLAP design.
III. SYSTEM FRAMEWORK AND IMPLEMENTATION Here we would mention the IBM BDW solution [3]. It is a
complete solution template for banking industry. Our template,
In this section, we introduce the framework and implemen- however, is tuned in some aspects such as organization and
tation of our system. human resource especially for the state owned large banks of
China.
A. Data Collection Process In the next two subsections, we detail the methods of two
There are multiple sources of data in the bank operation subjects, customer analysis and branch location analysis. Then,
process. We collect the data in the following ways: in the next section, we demonstrate their example application
in a real banking analysis.
• Flat files, including Excel files - mainly static data such
as branch information, staff infonnation, etc. D. Customer Analysis
• Relational database - often highly dynamic and huge In banking business, to better know the customers is a key
amount data, such as transaction data. factor to success. Many banks or even the branches of bank

144

Authorized licensed use limited to: GOVERNMENT COLLEGE OF TECHNOLOGY. Downloaded on February 27, 2009 at 09:44 from IEEE Xplore. Restrictions apply.
E. Branch Analysis
Most Chinese banks organizes by regional supervising bank
and elemental branches. As other retail industry, channel
construction becomes more and more important to gain better
positions in competition. Because branches plays so vital a
role in the retail bank, we put forward branch analysis in the
subject template.
Geographic location is still crucial although other channels
like e-bank emerges. To deeply understand the impact of loca-
tion, operational data alone is not enough. Actually, we have
another solution here which incorporate essential geographical
Fig. 2. Examples of subject template designed for banking intelligence. data to handle the problem [4]. The detail of the solution is
beyond the topic of this paper.

IV. EXAMPLE OF ApPLICATIONS


YEAR (ANAL_DATE)
I yearmonth In this section, we introduce a real-world application of our
morthname
yearmonthdate system to show its advantage in banking intelligence with more
DAY (ANAL_DATE)
details.
The data we are dealing with in this application comes from
a branch of state-owned Chinese bank. The branch locates in
a middle scale city in East China. We collected all the VIP
accounts as well as their transaction records in the past year
and fed the data to our solution.
Other than the basic analysis on age, education level
and etc., we carried through a combination analysis, which
revealed the transaction amount relating to age group and
transaction type simultaneously. Figure 4 illustrates the result.
From the figure, we can see that VIP customers of age group
31-60 contribute the most of transaction amount, and transfer
account occupies most amount among all the transaction types
across all the age groups.

Age group

Fig. 3. Example of OLAP design based on the template.

has a huge amount of customer, while among whom the VIP


customers contribute the best of the overall profit. We believe
the effort to analyze those VIP customers is the most cost
effective. As a consequence, in our solution we are mainly
focused on such customers.
Another benefit of analyzing the VIP customers is that
their profiles are usually complete. This gives the great help
on thoroughly understand the customers. We propose the
following fundamental factors to be enrolled:
Transaction type

• Age
Fig. 4. Example of combination analysis on VIP customers.
• Address
• RFV, i.e., recency, frequency and value of their transac-
tion behavior Another example is related to the branch analysis. We
noticed that quite a few customers opens account in one branch
In fact, the analysis includes but not limited to the above while handling business in other branches. We defined his most
factors. The OLAP tools gives the solution flexibility and visited branches as "favorite branch". Figure 5 shows that a
convenience to quickly employ the factors or any combinations great portion of VIP customers has a favorite branch different
of the factors. from the branch he opened the account. To give the proper

145

Authorized licensed use limited to: GOVERNMENT COLLEGE OF TECHNOLOGY. Downloaded on February 27, 2009 at 09:44 from IEEE Xplore. Restrictions apply.
explanation, we need to integrate the mentioned solution of
ours that takes the geographic information.

Branch name

Fig. 5. "Favorite branch" analysis.

v. CONCLUSION

In this paper, we introduce the importance and application


of banking intelligence in the modem bank industries. OLAP
lies in the heart of banking intelligence, and the method of our
solution includes a data warehouse architecture and subject-
oriented templates. We also demonstrate real-world application
results which implement the banking intelligence solution. The
findings may provide operators deep insight of various aspects
of operation.
REFERENCES

[1] M. Xie and W. 1. Yin, and J. Dong, Customer Profiling and Mining
with Multi-Dimensional Statistical Analysis: The Case of Retail Banking
Industry. JSM, 2007.
[2] S. Mansmann and T. Neumuth and M. H. Scholl, A OIAP Technology
for Business Process Intelligence: Challenges and Solutions, Data
Warehousing and Knowledge Discovery, Volume 4654, 2007.
[3] mM Whitepaper, IBM's Banking Data Warehouse and Basel II.
http://www.docstoc.comldocs/450950/IBMs-Banking-Data-Warehouse-
and-Basel-II.
[4] W. 1. Yin, and M. Xie and 1. Dong and B. Zhang, Mining Spatial Patterns
of Best Performing Branches in Banking Branch Marketing Analysis.
JSM, 2007.

146

Authorized licensed use limited to: GOVERNMENT COLLEGE OF TECHNOLOGY. Downloaded on February 27, 2009 at 09:44 from IEEE Xplore. Restrictions apply.

You might also like