CREDIT CARD FRAUD DETECTION USING BIG DATA FRAMEWORK
ABSTARCT : : In this paper, we are implementing a credit card fraud detection system, by using big data technologies. Credit card is one of the most divisive products among all the financial tools available.
The usage of credit cards has become common in today’s world and huge volume of transaction happens online. The increase in these transactions has also come with many apprehensions on the authenticity of the transactions. In today’s world, there have been various phishing attacks over the internet.
This needs to be dealt with caution. Be focus on designing an online credit card fraud detection framework with Big data technology. To accomplish that, we propose a workflow which satisfies most design ideas of current credit card fraud detection systems.
We implement it with largest Big data technologies like Hadoop, Spark, Apache Kafka etc. A prototype is implemented and tested with a synthetic dataset, which shows great potentials of achieving the above goals.
Due to the rapid emergence and evolution of online transactions, credit cards have become the most popular payment method. Credit card fraud involves using fake credit cards to purchase goods without paying.
On the other hand, researchers have proposed a wide range of anti-fraud systems with quick research and development around information technology and data mining including neural networks and decision trees, to advanced machine learning and deep learning methods.
Now -a-days, most of them are using credit cards for buying the goods which are so much in need but can't afford at the moment. In order to meet the needs credit cards are used and the fraud associated with it is also increasing so there is a need of developing a model that's fit well and predicts at higher accuracy.
Application fraud: When a fraudster acquires the control over the application, steals the credentials of customer, and makes a fake account and then the transactions takes place.
Electronic or manual card imprints: In this form of fraud, the fraudster skims the information from the magnetic strip which is present on the card then uses the credentials and fraud transactions are carried out
Card not present: This is a type of credit card in which physical card is not present during transaction
We propose a credit card fraud detection workflow, which can fuse different detection models to improve accuracy. It contains most of the common design ideas in latest CCFDSs, which make it much easier to integrate detection algorithms into the workflow.
Credit card transaction can be captured over time based on the behaviour of the cardholder’s spending incidents. The events are stored as transaction history which would be used to analyse the possible detection of fraud.
We implement the framework with the latest big data technologies like spark, kafka.we make use of cloud services for the storage of all the data. With these technologies, we are able to handle the burst amount of data and build a scalable and reliable system. Experimental results show that this system has the potential to achieve a sustainable performance.
We try to fuse existing algorithm, to achieve the goal, instead of developing a new algorithm. Here we are using hidden markov model (HMM) to detect fraud transactions.
Hidden markov model is perfect solution for detecting fraud transaction. The most important benefit of HMM is reduction in the number of false positive alarm i.e., considering a genuine transaction as a fraud transaction. Using HMM the cardholder’s spending behaviour can be observed to detect fraud.
Every arriving transaction is stored in the fraud detection system. The HMM model stores the transaction datasets in clusters, depending on the type of transaction. These clusters will be hidden from the outside world. Different types of transaction will be stored in different clusters.
|