Fraud Detection in Banking Sector: Using Big Data for Business Intelligence COIT 20253

This report discusses the methods of data storage and collection for significant data application at banks regarding fraudulent activities. It analyzes the data in action keeping recommendation system and customer-centric product design in mind. It also discusses how Big Data can be used for business continuity in case of power outrages and disasters. COIT 20253

0
FRAUD DETECTION IN BANKING SECTOR
COIT 20253
Report on
Fraud Detection in Banking Sector
Course Name: Business Intelligence
Using Big Data
Lecturer: Chandana Watagodakumbura
Tutor: Chandana Watagodakumbura
Submitted by
Pravallika Mallem,
12043395,
Melbourne Campus.

1
FRAUD DETECTION IN BANKING SECTOR
Executive summary
The report demonstrates the methods of data storage and collection for significant data
application at bank regarding fraudulent activities. Big data of the current era is highly
useful to find out the fraud and risks. The study analyzes the data in action keeping
recommendation system and customer-centric product design in mind. At last, a
discussion is made about how the online business of today can survive the power
outrages and disasters.
Table of Contents
Introduction:.............................................................................................................................2
Task 1: Data Collection and Storage through Big Data:.........................................................2
1.1. Data collection system:................................................................................................2
1.2. Storage systems:..........................................................................................................3
Task2. Data in action for banks:.............................................................................................4
2.1. Consumer-centric product design:...............................................................................4
2.2. Recommendation system:............................................................................................5
Task 3. Business continuityfor banks using Big Data:............................................................7
Conclusion:..............................................................................................................................8
Recommendation:...................................................................................................................9
References:...........................................................................................................................10

2
FRAUD DETECTION IN BANKING SECTOR
Introduction:
Banks of the current era has been increasingly turning to analytics for predicting
and preventing fraud at real-time. It has been at many times inconvenience for their
clients who have been making massive purchases or undergoing travelling. However,
the critical trouble for the banks has been top decrease billions of losses because of
that fraud.
Fraud detection at banks has been already looking at various factors like poor IP
addresses and unusual login times. On the other hand, Big Data application has been
drastically altering the approach with an enhanced analytic solution (Sobolevsky et al.
2014). They are fast and powerful enough for detecting fraud in real time and identifying
risks proactively.
The following report analyzes the data storage and collection for the current
scenario. Then it demonstrates the data in action regarding consumer-centric product
design and a recommendation system. Lastly, the business continuity is discussed
regarding how today’s online business survives the disasters and power outrages.
Task 1: Data Collection and Storage through Big Data:
1.1. Data collection system:
There are various kinds of data to be collected by banks through multiple processed.
The examples are encountered below.
The data to be
collected by banks
The processes
Corrupted data First of all the customers are to be found who have been
appearing on treasury department of forensic assent control.
Then the financial Action Taskforce must be ensured over
money laundering compliance. Then a list of transactions
must be produced with various agencies on the list of
different non-cooperative territories and countries.
Data about cash The transactions of money are to be identified below
regulatory thresholds of reporting. Then a series of cash
disbursements are to be determined through customer
number that exceeds the statutory reporting thresholds
together. Then the statistically unusual amount of transfer of
cash is to identify done by customers or through bank
accounts (Chen, Mao and Liu 2014).
Billing data They are retrieved through a huge number of waved fees
that are transferred to bank account and customer.
Checking tampering The out of sequence, void, duplicate and missing check
numbers are to be identified. Then checks paid are to be
determined that has never been matching checks that are

3
FRAUD DETECTION IN BANKING SECTOR
issued by check though banks. Then the falsification and
check forgery are to be located on various loan applications
(Martin 2015).
Skimming First of all the concise time deposits must be highlighted and
the withdrawal must be made on the similar account balance.
Then the indicators of the kiting checks are to be indications.
The duplication of skimming and credit card transactions are
to be highlighted.
Larceny First of all the customer account takeover must be identified
along with co-opted information about customer account
(Hilbert 2016). Then the number of loans must be located by
the bank employee or customer instead of repayments. Later
it must be found that the amount of investment has been
higher that value of a particular item or the collateral. Then
the sudden activities are to highlighted under dormant
customer accounts. This involves the identification of the
person which has been processing transactions against
those accounts. Lastly, the mortgage fraud schemes are to
be isolated through determining straw buyer indicators of a
plan.
Financial statement
fraud
For this, the suspense and dormant general ledger accounts
must be monitored. Then the journal entries are to be
identified at suspicious times.
1.2. Storage systems:
To understand the storage systems by big data, various factors are considered.
The first one is the area where the fraud takes place. The next one is the fraudulent
activities that have been looking like the data. The last one is the data source that is
needed to test the indicators of fraud.
Requirements for
storage
Methods to achieve them
Calculating the
statistical parameters
These include high/low values, standard deviations and
averages. The calculation is done by identifying outliers
indicating fraud.
Classification This is done through finding patterns among various data
elements.
Stratification of
numbers
These are done through identifying unusual entries like
excessively low or high values (Kitchin 2014).
Digital analysis
through Benford’s law
This is done through identifying various occurrences of digits
that have been naturally taking place at data sets.
Making joints with
various diverse
sources
This is done by Big Data through identifying matching values
like account numbers, addresses and names where they
must not exist.
Duplicating testing This is done through identifying different duplicate
transactions like expenses claims and payment of report

4
FRAUD DETECTION IN BANKING SECTOR
items.
Gap testing This is done through identifying various duplicate
transactions such as export report items claims and
payments.
Adding up of numeric
values
This is done through identifying various control totals which
can be falsified.
Validating entry dates This is to identify and inappropriate times to post or data
entry.
It must be noted that for storing data, different random samplings are not listed to
been smart fraud detection technique foe banks. Sampling has been an effective data
analysis technique to investigate data values. They have been consistent across the
population data. Here the very nature of fraud has been different since it has been
tending to take place randomly (Sobolevsky et al. 2015).
Task2. Data in action for banks:
Investigation of big data is the method to examine various huge data sets. This is
to uncover various patterns that are hidden. This involves the show changes in due time
and confirming and challenging theories. Very often they are relayed with data analytics,
decision-making and have applications beyond the for-profit world of business. Here,
the two methods consumer-centric product design and recommendation system are
investigated.
2.1. Consumer-centric product design:
It is also known as user-centric design. It is the process to frame services and
products across the various limitations and necessities of end users done by Big Data. It
has both regarding designing and quality of content, service and product (Varian 2014).
For a bank, it indicates the entire development and design process must take place
keeping the user of a database in mind from the step one. The various rules of
customer-centric design involve the following.
Filling the reservoirs of bank’s customer knowledge:
For instance, the salespeople of a bank are accountable to send enquiries of
customers on to the support. Here, the support staffs have been responsible for
funneling the contents of various customer conversations for the proper produce
person. Here, the project managers are accountable to decide what feedbacks must be
acted upon (Chen and Lin 2014).
Using customer knowledge to make the banks customer-centric:
Here the customer-centric product team are attentive to the voice of customers at
the early stage of product development prior its development, during construction and
after the expansion. Next, the customer experiences and needs ate considerable factors
to undertake decisions regarding what latest features have been implemented, where to
expand and push the vision forward (Cao 2015).

5
FRAUD DETECTION IN BANKING SECTOR
Prioritizing roadmaps for business and users:
This can be effectively done through using the formula (Value/Complexity)*Necessity^2.
This must be done regularly to weigh the long listing ideas and to pick the one
that is needed to be given the most priority. Here for all the ideas, a score from 1 to 10
must be assigned for every variable. Next, the totals score must be calculated ranking
every idea.
Solving issues for the user during the creation:
Here, the critical phase has been designing the user experience. As the
implementation process goes on, here, the customer-centric team focuses on
developing and delivering a user experience that has been easy and intuitive for
customers to use (Hu et al. 2014). Further, the design of user experience has been the
design reframed. This term emphasizes solving problems for users. This is the reminder
that the model has not been making pretty. Besides, it is the process to craft the holistic
beginning-to-end experience accounting every detail of journey of users.
Communicating changes to a customer after the development is complete:
As the implementation is done, a new feature request is needed to be made. This
is intended to be accountable regarding communicating changes to customers. Here,
the communication has been the important method of establishing transparency and
trust regarding delivering. It must be reminded for banks that quick turnarounds over
bank’s customer feature requests have been in a long way to go. However, this has
been a slower turnaround that must garner till there is any communication (Stimmel
2016).
2.2. Recommendation system:
The recommended systems for banks are complicated approaches to achieve
customer goals. The various steps undertaken by Big Data analysis are analyzed
below.
Problem facing:
It includes the collection of expectations, requirements and issues from
concerned employees and business departments. Here, the interviews are conducted
with stakeholders with various clarified needs for the system and managing vision of
clients.
Data processing and collection:
At this step, every necessary CRM and various transactional data for the clients
of banks are needed for subsequent data analysis and preprocessing. Here, the data
set includes social-demographical data, purchase history of clients, history of balances,
monthly payments and information about the client. The step involves the checking of
data consistency taking place among various sources and treatment of inconsistencies.

6
FRAUD DETECTION IN BANKING SECTOR
It further includes the handling of duplicates and null values (Li, Cao and Yao 2015).
Moreover, this provides dictionary transformation and forming along with statistics
calculation. Here, the problems detected and various solutions to it are reported to
clients.
Data processing and development of algorithms:
This stage starts with data augmentation actions that result in mining changes to
spend speed or growth of income and various client event triggers. For understanding
the financial behavior, the banks must calculate the aggregate balances and outcome or
income transfers for multiple periods. Here the features are merged with different
elements from CRM system like purchase history of clients, social-demographic data
and so on (Olson and Wu 2017). Here, various types of machine algorithms are
applicable on the features to predict best recommendations regarding products for
clients.
A/B testing:
For proofing the effectiveness of designed solution the banks must plan an A/B
test. Here, there are three groups for the experiment. The first one is the group formed
by various segmentations which are done as per social and demography. This also
includes different reference groups without any recommendations. The banks have
been offering credit cards and multiple cards through call-downs at call centers,
premium and debit tickets through SMS broadcasting. This is for various groups of
clients (Kim, Trimi and Chung 2014). Outcomes of A/B testing display significant
excellence according to sales and conversions. These are done in the groups created
by recommender algorithm as compared to other groups.
25%
12%
25%
38%
Expanding Deploying Experimenting Exploring
Figure 1: “Various Levels of Big Data maturity and adoption in banks”

7
FRAUD DETECTION IN BANKING SECTOR
(Source: Mulder et al. 2016, Page no.2053951716662054)
Task3. Business continuity for banks using Big Data:
In this section, how banks survive the case of power outrage and various other
disasters are discussed. It is seen from various instances, that various vulnerabilities
within power systems have been affecting various regions. It has shown different real-
world instances regarding how slow the power companies have been taking to react as
the disaster takes place. A study conducted on 2015 revealed that some of those
inconsistencies and vulnerabilities are present in various banking systems. It proved
that there has been delayed and prolonged recovery plan in that failure event. It was
seen that the information has been inaccurate and has been taking about ten days in
restoring power to the coverage area. As known today, the study has much more
accurate (Thorstad and Wolff 2018).
Figure 2: “Opportunities for Big Data in Banking Risk Management”
(Source: Verhoef, Kooge and Walk 2016)
As a result, the banks of today have been working on big data systems helping
alleviate some of those issues for the future. Hence any analytics solution is needed to
be designed for assisting companies in managing power outrages and recovering
faster. In the next phase, predictive analysis is required to be handled for allowing
officials for reacting as per in the event they need to plan for outage or system failure.
For example, the famous software named PowerON response can be used by
banks. This has been designed for collecting various data from network and field

8
FRAUD DETECTION IN BANKING SECTOR
Unknown Till unknown Known
Signal
service crews. They can manage and assess data regarding facility damages, circuit
conditions and failures, equipment and devices and restore periods for customers.
Further, the data can also be shared with various sources that include management
teams, utility crews and even customers (Frizzo-Barker et al. 2016).
Here, the predictive side of the software has been obviously helping various
groups to make preparation for further scenarios and then identify the different weak
points within that system. Further, the field crews can react through changing the old
tools, repairing damages devices and retrofitting previous circuit breakers. Here, the aim
has been to help the banks and their teams avoid different power outrages and
additional outrages as much possible. Otherwise, it is intended to at least recover very
quickly as anything goes wrong (Hale and Lopez 2018).
Here, all the collected data and software must be sued to prevent the wide-scale
outages. This is helpful for people as they are left without electricity. However, here the
worst case is that it could be easily preventable and augmented with various factors.
These are the two most relevant factors are poor coordination and miscommunication
taking place between service teams. As any big data system is available, the scenario
gets easier differently (Meier 2015). And, this is the exact thing on which banks and its
power-related clients have been betting on. As it quells and prepares for various failures
prior it can happen, it assures that people are not left behind in the dark for long.

9
FRAUD DETECTION IN BANKING SECTOR
Figure 3:” Opportunities of Big Data in Banking Risk Management”
(Source: Meune et al. 2017, Page No. 38)
Conclusion:
Since more and more data are been generated and gathered through big data,
analytic capabilities are continuing to progress. Thus, banks can create real value
through monetization of data. Performing that accurately helps to maximize the value
and secure every vital corporate reputation and brand. In conclusion, it can be said that
the potential and promise of significant data needs are needed to be matched though a
considered approach to use, license, store and collect. Remedies regarding misuse of
data at bank are found to be challenging as there is devoid of any well though data
strategy. Further, current copyright protections can help, contract and undertake
confidential solutions for banks that have been more very much significant.
Recommendations:
The various suggestions to check fraudulent activities taking place at banks through big
data is identified below. Using data science solution pillars:
Banks can search data using Apache Solr to seek and create models. This can
be done by utilizing Apache Pig, Spark and ecosystem analytics tool like R. Machine
learning models. These can be tested and various investigations could be run on real
and substantial data sets for improving and iterating processing. Real time recommended offers:
New product and services are to be targeted for right customers through
implementing analytics engine supporting integrated and flexible process. This is done
by a better understanding of motivations, buying patterns, preferences and customer
needs. Making proper compliance with laws and regulations:
Various rules and regulations apply to several types of data and affect the
security, sales, uses, storage and collection. Every applicable laws and regulation are
needed to be ensured. These must be addressed as per the type of data while issuing.

10
FRAUD DETECTION IN BANKING SECTOR
References:
Cao, L., 2015, ’Big data analytics-innovation and practices’, In Reliability, Infocom
Technologies and Optimization (ICRITO)(Trends and Future Directions), 2015 4th
International Conference, pp. 1-1, viewed 7 April 2018,
http://www.amity.edu/aiit/icrito2015/ICRITO'2015%20Sessions.pdf
Chen, M, Mao, S & Liu, Y 2014, ‘Big data: A survey’, Mobile Networks and
Applications, vol.19, no.2, pp.171-209, viewed 3 April 2018,
https://dl.acm.org/citation.cfm?id=2843712
Chen, X.W & Lin, X 2014, ‘Big data deep learning: challenges and perspectives’, IEEE,
vol.2, pp.514-525, viewed 4 April 2018,
https://pdfs.semanticscholar.org/2724/5e65a27bde90b5b0bb25d157bb75a0ad8b5a.pdf
Ekbia et.al, 2015, ‘Big data, bigger dilemmas: A critical review’, Journal of the
Association for Information Science and Technology, vol.66, no.8, pp.1523-1545,
viewed 4 April 2018,
https://onlinelibrary.wiley.com/doi/pdf/10.1002/asi.23294
Frizzo-Barker et.al, 2016, ‘An empirical study of the rise of big data in business
scholarship’, International Journal of Information Management, vol.36, no.3, pp.403-
413, viewed 4 April 2018,
http://daneshyari.com/article/preview/1025514.pdf
Hale, G & Lopez, J.A 2018, Monitoring Banking System Fragility with Big Data, Federal
Reserve Bank of San Francisco.
Hilbert, M, 2016, ‘Big data for development: A review of promises and challenges’,
Development Policy Review, vol.34, no.1, pp.135-174, viewed 5 April 2018,
https://onlinelibrary.wiley.com/doi/abs/10.1111/dpr.12142
Hu et.al, 2014, ‘Toward scalable systems for big data analytics: A technology
tutorial’, IEEE, vol.2, pp.652-687, viewed 5 April 2018,
https://ieeexplore.ieee.org/abstract/document/6842585/
Kim, G.H, Trimi, S & Chung, J.H 2014, ‘Big-data applications in the government
sector’, Communications of the ACM, vol.57, no.3, pp.78-85, viewed 6 April 2018,
https://dl.acm.org/citation.cfm?id=2500873

11
FRAUD DETECTION IN BANKING SECTOR
Kitchin, R 2014,‘The real-time city? Big data and smart urbanism’, Geo Journal, vol.79,
no.1, pp.1-14, viewed 6 April 2018,
https://pdfs.semanticscholar.org/6e73/7a0e5ef29303760a565ba5e9d98510ab0976.pdf
Li, D, Cao, J & Yao, Y 2015, ‘Big data in smart cities’, Science China Information
Sciences, vol. 58, no.10, pp.1-12, viewed 6 April 2018,
http://epublications.bond.edu.au/cgi/viewcontent.cgi?article=1478&context=fsd_papers
Martin, K.E., 2015 Ethical issues in the big data industry, Browser Download This
Paper.
Meier, P 2015, Digital humanitarians: how big data is changing the face of humanitarian
response. Crc Press, viewed 5 April 2018,
https://www.crcpress.com/Digital-Humanitarians-How-Big-Data-Is-Changing-the-Face-
of-Humanitarian/Meier/p/book/9781482248395
Meune, C 2017, ‘P4895Current use and misuse of troponin measurements form a large
cohort: results of big data analysis’. European Heart Journal, vol.38, no.1, viewed 5
April 2018,
https://academic.oup.com/eurheartj/article/38/suppl_1/ehx493.P4895/4086414
Mulder, F 2016, ‘Questioning Big Data: Crowd sourcing crisis data towards an inclusive
humanitarian response’, Big Data & Society, vol.3, no.2, viewed 5 April 2018,
https://doi.org/10.1177/2053951716662054
Olson, DL & Wu, D 2017, ‘Predictive Models and Big Data’, In Predictive Data Mining
Models , pp. 95-97, Springer, Singapore.
Reimsbach-Kounatze, C 2015, ‘The Proliferation of “Big Data” and Implications for
Official Statistics and Statistical Agencies’, viewed 6 April 2018,
https://www.oecd-ilibrary.org/docserver/5js7t9wqzvg8-en.pdf?
expires=1523874446&id=id&accname=guest&checksum=32D97D4990081385D34AA2
161A42F89E
Sobolevsky, S 2015, ‘Scaling of city attractiveness for foreign visitors through big data
of human economical and social media activity,In Big Data (Big Data Congress)’, IEEE
International Congress, pp.600-607, viewed 6 April 2018,
https://arxiv.org/ftp/arxiv/papers/1609/1609.09340.pdf
Sobolevsky, S 2015, ‘Money on the move: Big data of bank card transactions as the
new proxy for human mobility patterns and regional delineation,In Big Data (Big Data
Congress)’, IEEE International Congress, pp. 136-143, viewed 6 April 2018,
https://dl.acm.org/citation.cfm?id=2671976

12
FRAUD DETECTION IN BANKING SECTOR
Stimmel, CL 2016, Big data analytics strategies for the smart grid, Auerbach
Publications.
Thorstad, R & Wolff, P 2018, ‘A big data analysis of the relationship between future
thinking and decision-making’, Proceedings of the National Academy of
Sciences, vol.115, no.8, pp.E1740-E1748, viewed 8 april 2018,
https://doi.org/10.1073/pnas.1706589115
Varian, HR 2014, Big data: ‘New tricks for econometrics’. Journal of Economic
Perspectives, vol.28, no.2, pp.3-28, viewed 9 April 2018,
https://www.aeaweb.org/articles?id=10.1257/jep.28.2.3
Verhoef, PC, Kooge, E & Walk, N 2016, Creating value with big data analytics: Making
smarter marketing decisions, Routledge.
Zhong, RY 2016, ‘Big Data for supply chain management in the service and
manufacturing sectors: Challenges, opportunities, and future perspectives’, Computers
& Industrial Engineering, vol.101, pp.572-591, viewed 10 April 2018,
http://dx.doi.org/10.1016/j.cie.2016.07.013