Irjet V5i3669 PDF
Irjet V5i3669 PDF
Irjet V5i3669 PDF
1. INTRODUCTION
In era of plastic cards with rapid advancement of
electronic commerce, the credit card has become
convenient and de facto standard for online shopping. The
increased number of credit card transaction open the door
for thieves to steal credit card details and commit fraud. it
impacts a fraction of percent of all purchases made with
plastic, according to data from Federal Reserve ,it
represents one of the biggest concerns among customers.
Due to this card issuers bore a 63% share of fraudulent
losses in 2012 and merchants assumed the other 37% of
liability, according to the Nilson Report, August 2013. So it Fig -1: Overall flow of the algorithm
becomes essential to improve the fraud detection system
to minimize the losses. The credit card fraud detection 3. Experiment Process
system presents a number of challenging issues for data
mining. The past transactions’ data of the user is collected from
the bank’s database. The current transaction’s data is
This paper is to purpose a credit card fraud detection compared against the patterns generated from the
system using outlier analysis. Outlier is a data point which previously collected data. To generate the patterns based
is significantly different from the remaining data and upon the user’s behavior, various fraud detection rules are
deviates so much from other observation as to arouse considered. Every user is provided with a score attribute
suspicions that it was generated by a different which is stored in bank’s database. This score is
mechanism[10]. Outlier analysis of transactional data incremented if the current transaction’s data of the user
depending on its past behavioural patterns is done to label passes the rule and is decremented otherwise. If the
transaction as either fraudulent or not. For analysis, Rule current transaction’s data of the user fails any rule or his
Engine is implemented. current score goes below the threshold, then the
transaction is suspicious.
© 2018, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 2921
International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 05 Issue: 03 | Mar-2018 www.irjet.net p-ISSN: 2395-0072
Step 3 : The current transaction’s data is tested across Table -1: Rules in Rules Engine
all the rules and score is updated accordingly.
Rule Name Pass Condition Fail Condition
Step 4 : The score is checked against the threshold and
transaction is labeled as suspicious or normal transaction. User Authentication Valid credentials Invalid credentials
© 2018, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 2922
International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 05 Issue: 03 | Mar-2018 www.irjet.net p-ISSN: 2395-0072
1. Card number 5500000000000004 has made previous [4] Wen-Fang YU, Na Wang-Research on Credit Card
transactions at locations with latitude and longitude Fraud Detection Model Based on Distance Sum, IEEE
values as [ (17.532, 21.576), (18.345, 22.534), International Joint Conference on Artificial
(17.655, 22.654), (16.651, 20.127), (20.764, 72.765), Intelligence 2009.
(21.64, 71.75), (20.94, 73.765) ]. The current
[5] Amlan Kundu, Suvasini Panigrahi, Shamik Sural and
transaction is made from location (100.654, 181.409).
Arun K. Majumdar, ―BLAST-SSAHA Hybridization for
The clusters formed are [17.532, 21.576), (18.345,
Credit Card Fraud Detection,‖ IEEE Transactions On
22.534), (17.655, 22.654), (16.651, 20.127) ] and
Dependable And Se-cure Computing, vol. 6, Issue no. 4,
[(20.764, 72.765), (21.64, 71.75), (20.94, 73.765)].
pp.309-315, October-December 2009.
The current transaction’s data doesn’t fit into any one
of the cluster’s radius range. Hence the score is [6] Ashish Gupta 1, Jagdish Raikwal- Fraud Detection
decremented as the transaction is suspicious. credit card Transaction using Hybrid model-
International Journal of engineering and computer
2. Card number 4111111111111111 has made previous science ISSN:2319-7242 Volume 3 Issue 1 Jan, 2014.
transactions of amounts [1000, 1500, 1200, 500,
2000, 15000, 17000, 20000]. The current transaction [7] K.RamaKalyani, D.UmaDevi - Fraud Detection of Credit
is made of amount 1,000,000. The clusters formed are Card Payment System by Genetic Algorithm,
[ 1000, 1500, 1200, 500, 2000 ] and [ 15000, 17000, International Journal of Scientific & Engineering
20000 ]. The current transaction’s data doesn’t fit into Research Volume 3, Issue 7, July-2012.
any one of the cluster’s radius range. Hence the score
is decremented as the transaction is suspicious. [8] Wang Xi- Some Ideas about Credit Card Fraud
Prediction China Trial. Apr. 2008.
5. CONCLUSION
[9] White paper- How a hybrid Anti-Fraud approach
This paper presents mechanism of credit card fraud could have saved government benefit programs more
detection and examines the result based on the principles than $100 millions.
of clustering algorithm. Using outlier analysis this paper,
tries to minimize false alerts and improving existing
models for detection of fraud. In this study fraud detected
and fraud transactions are generated by maintaining the
fraudulent history table. If this is applied into bank credit
card fraud detection system, the probability of fraud
transactions can be predicted soon after credit card
transactions by the banks. Though fraud in online card
payment cannot be eradicated still this study is trying to
minimize fraud.
ACKNOWLEDGEMENT
We render our sincere thanks to Barclays team (BTCI,
Barclays Technology Centre India) for their support and
encouragement.
© 2018, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 2923