Sei sulla pagina 1di 5

CREDIT CARD FRAUD DETECTION

Dr.Vishal Shrivastava Anjali Dharwal


M.tech Co-ordinator, Dept. of CS/IT M.tech Scholar, Dept. of CS/IT
Arya College of Engineering & I.T. Jaipur Arya College of Engineering & I.T. Jaipur
vishalshrivastava.cs@aryacollege.in anjidharwal17@gmail.com

ABSTARCT-Payment mode accepted both the fraud from the huge data set where lots of
offline and online is credit card in cashless legal transactions have done. There are very few
transaction. It is convenient and easy to make paper on credit card fraud detection methods due
payments and other transactions. Credit card to the fact that these methods can be tested
fraud is also growing along with the without a dataset. So it’s difficult to prove the
development in technology. It can also be said probability of success ratio of the methods. As
that economic fraud is increasing in the We know that the credit card information is
global communication improvement. It is confidential, the bank owners and service
being recorded every year that the loss due to Providers do not encourage in sharing these with
these fraudulent acts is billions of dollars. All anyone.
banks have need to minimize confusion and In everyday life we are using credit card
fraud by using technology and efficient transactions to make our life easy and
methods. There are several new and old convenient. Credit card transactions are evolving
techniques like Machine learning, Genetic to a world of effortless expenditure. Frauds
Programming, fuzzy logic, sequence involved in the credit cardsection have caused
alignment, etc are used for detecting credit severe damage to the users and the service
card fraudulent transactions from many provider and is said to be even worse in coming
years. Along with these techniques, these days.
approaches are proved to minimize the false Data mining, the extraction of hidden predictive
alarm rates and increase the fraud detection information from large databases, is a powerful
rate. Any of these methods can be new technology with great potential to help
implemented on bank credit card fraud companies focus on the most important
detection system, to detect the false alarm and information in their data warehouses. Data
prevent the fraud detection in transactions. mining tools predict future trends and behaviors,
allowing businesses to make proactive,
Keywords: Fraud Detection, technology, transaction, knowledge-driven decisions. The automated,
Outlier Detection. prospective analyses offered by data mining
move beyond the analyses of past events
I. INTRODUCTION provided by retrospective tools typical of
decision support systems.
Being ecommerce merchant, the major problem
faced due to these fraud activities is so similar to
legal ones. Complex fraud detection must be
detected. The major challenging part to detect
inthe HMM. The author has suggested method
for finding the spending profile of cardholders.
It is also discussed how the HMM can identify
the fraudulent transactions. The simulation
results show the advantages of using HMM and
learning the profile of the cardholder plays an
important role in analyzing fraudulent cases. The
result also shows that 80% of the results are
FIG 1. accurate and the system is scalable for large data
set as well.
II. LITERATURE REVIEW
III. PROPOSED WORK
Divya.Iyer et al [1] the author uses Hidden In day-to-day usage of credit card transactions
Markov Model (HMM) to detect credit the procurement of products and services assists
cardtransaction frauds. The training set is tuned online transactions or card swiping
with the normal behavior of the card holder. So procurements. This leads to increase in online
if credit card transaction is rejected by the transactions using credit and debit cards
trained HMM then that transaction is said to be evolving to a world of effortless expenditure.
fraudulent. Care is to be taken that valid and Frauds involved in the credit card section have
genuine transactions are not considered as fraud. caused severe damage to the users and the
The author also compares various methods with service provider and is said to be even worse in
the proposed methods to prove that HMM is coming days.
much preferred than the other methods.
There are several techniques like Machine
Renu et al K.RamaKalyani et al [2] creates a learning, Genetic Programming, fuzzy logic,
test data and through which the fraudulent sequence alignment, etc are used for detecting
activities are detected. This algorithm is also credit card fraudulent transactions. Along with
called as an optimization technique based on these techniques, outlier detection methods are
genetic and natural selection in high implemented to optimize the best solution for
computational problems. The author proposes a the fraud detection problem. These approaches
method to detect credit card fraud and the results are proved to minimize the false alarm rates and
are validated using principles of this algorithm. increase the fraud detection rate.
The purpose of detecting fraud cases is to
declare it to the client and the service provider. IV. Problem Formulation
K.RamaKalyani et al [3] proposed a fraud Frauds involved in the credit card section have
detection method which involves monitoring the caused severe damage to the users and the
activities of populations to observe and predict service provider and is said to be even worse in
undesirable behavior. Undesirable behavior is a coming days. Fraudsters observe and adapt to
set of several habits like intrusion, fraud, the quick changes in the technology and find
delinquency and defaulting. This research clever ways to involve in illegal activities.
speaks on several credit card fraud detection and Frauds caused due to these smart hackers are
telecommunication fraud and different hazardous and dangerous. A well-educated
techniques which help in resolving the discussed fraudster can create several identities and
problems. conduct credit card transactions without being
caught. Talking in terms of e-commerce
AbhinavSrivastava et al [4] the author uses the transactions the major problem faced due to
ranges of transaction amount as an attribute these fraudulent activities is so similar to legal
ones. Hence having an efficient and complex  It produces a simple probability
fraud detection system is a must to prevent these formula for classification.
fraudulent activities.  It works well with linear data
for credit card fraud detection.
 Statistical classification is  It cannot be applied on non-
performed on statistical linear data in data set.
characterizations of patterns which  It is not capable of handling
are generated using probabilistic fraud detection at the time of
system. credit card transaction.
 Whereas structural classification 2. Decision Tree-
depends on the structural  This method can handle non-
interrelationships of features. linear data as well.
 It involves complex
Supervised classifications are a process in which algorithm even small
we gather the information of interest from the and small structure of the tree.
dataset and consider it as training data. The Choosing splitting criteria isalso
statistical classification of each training set is complex.
obtained. After the statistical classification is 3. Support vector machine-
performed the best match from the training data  This method is capable of
is considered and the class it resembles is detecting the fraud activity at
chosen. the time of credit card
transaction.
V. Objectives  Sometimes it fails to detect
fraud alarms.
Credit card fraud detection is considered highly 4. K-Nearest Neighbor Algorithm-
confidential and not disclosed to anyone.  There is no requirement of
Popular payment mode accepted both offline predictive model
and online is credit card that provides cashless Before classification.
transaction.  The accuracy of the method
depends on themeasure of
Fraud detection is a highly complex function to distance.
be performed where there is no system which
can guarantee a 100% satisfaction result rate. All
the existing methods can likely predict fraud VII. MACHINE LEARNINGIN
transactions and not assure you about the results. FRAUD DETECTION
Let’s consider the properties of good fraud
detection method: Machines are much better than humans at
processing large datasets. They are able to detect
a) It must be able to identify the frauds and recognize thousands of patterns on a user’s
accurately. purchasing journey instead of the few captured
b) It must quickly detect fraud cases. by creating rules. We can predict fraud in a large
c) At any case a genuine transaction should volume of transactions by applying
not be considered as fraud. cognitivecomputing technologies to raw data.
This is the reason why we use machine learning
VI. Research Methodology algorithms for preventing fraud for our clients.
The three factors which explain the importance
of machine learning are –
1. Logistic Regression-
 Speed –In rule-based systems, In outlier detection method unsupervised learning
people create ad hoc rules to is preferred to detect the fraud because it can lead
determine which types oforders to to new explanations and representation of the
accept or reject. This process is observation data. Another
time-consuming and involves
manual interaction. As the velocity
of commerce is increasing, it’s very IX. CONLUSION
important to have a quicker solution
to detect fraud. Credit card scam has become much more
 Scale – Machine learning algorithms extensive. To progress safety measures of the
and models become more effective monetary transaction systems in a habitual and
with increasingdata sets. Whereas in effectual way, structure a precise and well
rule-based models the cost of organized credit card scam detection system is
maintaining the fraud detection one of the essential functions for money
system multiplies as customer base transactions.. By the mean time outlier detection
increases. mechanism helps to detect the credit card fraud
 Efficiency –Contrary to humans, using less memory and computation
machines can perform repetitive requirements. Especially outlier detection works
fast and well on online large datasets.
VIII. OUTLIER DETECTION
X. REFERENCES
An outlier approach is very different from
[1]AbhinavSrivastava, AmlanKundu,
the traditional observation method. Outlier ShamikSural, and Arun K. Majumdar” Credit
method is used to detect unusual behavior of Card Fraud Detection Using Hidden Markov
a system using a different mechanism. Model” VOL. 5, NO. 1,JANUARY-MARCH
Outlier detection is classified into two main 2008
types based on which labels can be assigned [2]K.RamaKalyani, D.UmaDevi” Fraud
to each group. Detection of Credit Card Payment System by
Genetic
In supervised outlier detection method domain [3]P.Jayant,Vaishali,D.Sharma,” Survey on
experts model the system to learn and Credit Card Fraud Detection Techniques”,
classifierthe outlier using training set of data. [4]VenkataRatnamGanji,” Credit card fraud
detection using Anti-k Nearest
In unsupervised outlier detection objects are Neighbor Algorithm”,International Journal on
clustered into multiple groups based on features. Computer Science andEngineering (IJCSE) Vol.
Those objects are far from any group is labeled 4 ,06 June
as outliers. [5]EkremDuman, M. HamdiOzcelik “Detecting
credit card fraud by geneticalgorithm and scatter
search”. Elsevier, Expert Systems with
Applications.

FIG:- 2

Potrebbero piacerti anche