International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395 -0056
Volume: 04 Issue: 03 | Mar -2017
p-ISSN: 2395-0072
www.irjet.net
Unsupervised Learning for Credit Card fraud detection Professor. Vikrant Agaskar1 , Megha Babariya2, Shruthi Chandran3 , Namrata Giri4 1Professor,
Department Of Computer Engineering, VCET, Mumbai University, India Department Of Computer Engineering, VCET, Mumbai University, India 3Student, Department Of Computer Engineering, VCET, Mumbai University, India 4Student, Department Of Computer Engineering, VCET, Mumbai University, India
2Student,
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Today, Internet banking has led to an increase of
a database. Here the ‘data stream analysis’ is done. The upcoming of new data is endless. Thus, effective algorithms must be able to work with a constant memory footprint, regardless of the evolution of the stream, as the whole database cannot be kept in memory.
frauds, resulting in substantial financial losses. Banking frauds increased 93% in 2009-2010, and 30% in 2012-2013. Internet banking frauds are difficult to analyze and detect because the fraudulent behavior is dynamic, spread across different customer’s profiles, and dispersed in large and highly imbalanced data sets. Customers do not check their banking history daily to analyze any kind of fraud. We propose in this paper a technique of synthetic model of the data structure for efficient storage of data, and a measure of dissimilarity between these representations for the detection of change in the stream structure , in order to detect different types of fraud during a period of time.
There are many issues that make effective fraud management a challenging task. These include: large and ever-expanding mass of data, the growing complicatedness of systems, changes in business processes and activities and continuous transformation of new fraud schemes to avoid the existing detection techniques. To detect the fraudulent financial statements is a difficult job when using normal audit procedures due to limitation in understanding the characteristics of financial statements, lack of experience and dynamically changing strategies of fraudsters.
Key Words: Clustering , fraud detection , Transaction , Unsupervised Learning , data mining algorithms.
Supervised methods, using samples from the fraudulent/non-fraudulent classes as the basis to construct classification rules to detect future cases of fraud, to prolong from the problem of unbalanced class amount: the legitimate transactions are more in number then the fraudulent ones.
1. INTRODUCTION In this digital age, it is the time of online banking, one of the well-organized and easier modes of transaction. With the evolution of internet in the banking sectors, people have changed the way they used to bank. But this digital transformation is providing new ways for fraudsters to hack people’s private accounts. Banking sector frauds have been in existence for centuries, with the earliest known frauds pertaining to insider trading, stock manipulation, accounting irregularity/ inflated assets etc. Fraud is a superior form of white collar crime that persist to extract a significant toll not only on the organizations, but also on investors, financial institutions, and the economy in general.
In this paper, we propose an unsupervised method for detecting fraudulent transactions using records of the amount and location details of previous transactions carried out by the customers.
2. RELATED WORK Construction of a synthetic model at daily intervals over a data repository which vacants itself as new data is stored. This synthetic representation is being derived from the learning of a weighted SOM (Self-Organizing Map) and admits automatic data clustering. During the learning procedure, each pattern is extended with novel information uprooted from the data.
Most IT entities today in use are transactional. This means that the data transactions are processed in the system and the data of transaction is stored in the system’s database. The relationships and patterns in these reserved transactional data is analyzed by the data mining softwares. One of the major issues during this is the fraud which may occur during transactions..
First, there is a shortage of information concerning the characteristics of management scam. Secondly, with the irregularity it has, most of the auditors need the necessary experience to detect it. Finally, managers intentionally try to mislead the auditors. These limitations demonstrate that there is a need for supplementary analytical procedures
Databases constantly keep on changing and the size of data keeps on increasing . This makes the transformation and mass of data so vital that it is impossible to store them in
© 2017, IRJET
|
Impact Factor value: 5.181
|
ISO 9001:2008 Certified Journal
| Page 2343