International Research Journal of Engineering and Technology (IRJET) Volume: 09 Issue: 05 | May 2022 www.irjet.net
e-ISSN: 2395-0056 p-ISSN: 2395-0072
E-Commerce Transaction Fraud Detection through Machine Learning Rajat Kumar1, Prajakta Jadhav2, Rishabh Yadav3, Sakshi Pawar4, Prajyot Yawalkar5, Nehali Shinde6 1,2,3,4,5,6 IT
dept, Dhole Patil College of Engineering, Pune -----------------------------------------------------------------------***--------------------------------------------------------------------Abstract — The number of transactions have been Such procedural and rule-based procedures have been
unsuccessful in past few years as scammers have figured out how to circumvent the inflexible regulations. Contrary to the most recent industry statistics, the amount of worldwide financial corruption is expected to skyrocket. Embezzlement is a big proportion of the unauthorized charges in this category, accounting for a considerable share of the frauds.
steadily increasing consistently over the past few years. The development of online financial services in the form of credit cards, online funds transfer and United Payments Interface or UPI have catalyzed the growth further which has led to the astronomical number of transactions. The number of fraudsters or scammers has also been increasing consistently, which are performing fraudulent transactions. There are numerous fraudulent transaction detection techniques that are put in place by the financial institutions but are unable to detect the ingenious frauds committed by the criminals. Therefore, this paper defines an effective approach for the purpose of fraudulent transaction detection through the use of Linear Clustering, Entropy Estimation and frequent itemset extraction along with Hypergraph formation, Artificial Neural Networks and Decision Making. The extensive evaluation has been performed for quantifying the approach which has resulted in the expected outcomes.
The expense of fighting and recovering from fraud has also risen significantly. In conjunction to the rise in fraudulent activity, fraud tactics have evolved significantly. Card embezzlement has decreased dramatically in recent years as digital transactions have been more widely used. However, there has been a significant increase in potential digital transaction fraud. Due to the worldwide epidemic and the resulting substantial surge in digital trade volumes, electronic fraudulent credit card incidents increased dramatically. Credit card transaction is becoming one of the most popular means of payment in recent years, thanks to the fast advancement of digital transaction. Nevertheless, the advent of contactless electronic monetary operations has resulted in the introduction of new forms of financial fraud. Scam artists frequently collect information and data from investors in order to conduct illicit transactions in a brief span of time. As a result, banking firms should employ a variety of approaches in the physical world and in internet to strengthen credit card fraud monitoring and defend consumers' security.
Keywords: Artificial Neural Network, Information gain, Hyper graph, neo4j, transaction Fraud.
I. INTRODUCTION In the last decade, digital payments have grown at an unprecedented rate. Overall transaction amounts are up over last year, and transfer volumes have increased significantly. Correspondingly, in India, the amount of electronic transactions has increased dramatically in recent years. Internet banking as well as mobile payment have given billions of individual’s access to banking services throughout the world. They have also brought tangible opportunities to consumers, organizations, and financial intermediaries, including as the potential to expand, lower operational costs, simplicity of use, convenience, and improved efficiency.
These cases of frauds are getting increasingly sophisticated and are highly problematic as the current approaches have been insufficient in the detection and identification of the frauds. This has become an increasingly difficult to detect as the static conditions are easily identified and circumvented by these scammers. The development of counter-measures is extremely speedy and these techniques can remain undetected for a long period of time allowing the criminals to perpetuate their crime. Therefore, there is a need for an effective and useful approach for the purpose of achieving the fraudulent transaction detection. The paradigm of machine learning techniques has been one of the most significant for the purpose of transaction fraud detection.
Illegal strategies, on the other side, have quickly adapted to take advantage of the new fast-paced electronic payments scenario. Historically, embezzlement and economic fraud identification depended on a vast number of regulations and fixed criteria, such as maximum transaction restrictions, to identify questionable activities. © 2022, IRJET
|
Impact Factor value: 7.529
|
ISO 9001:2008 Certified Journal
|
Page 1273