International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 11 Issue: 04 | Apr 2024
p-ISSN: 2395-0072
www.irjet.net
Online Transaction Fraud Detection Using Machine Learning Dr.Ranjit K N1, Ms. Bhoomika Rajendra Vernekar2, Ms. Chandana MR3, Ms. Spandana MP4, Mr. Mallikarjun Bachwar5 1 Associate Professor, Dept. of Computer Science and Engineering, Maharaja Institute of Technology,
Thandavapura
2,3,4,5Students, Dept of Computer Science and Engineering, Maharaja Institute of Technology, Thandavapura
---------------------------------------------------------------------***--------------------------------------------------------------------payment requests to select which transactions to authorize. Abstract – Nowadays, people conduct practically all their
Every authorized transaction is examined, and any suspicious activity is found, using machine learning algorithms. Before chatting with the experts looking into these claims, cardholders need to determine whether a transaction was legitimate or fraudulent. The automated system that creates and refines the algorithm incorporates suggestions from the detectives to gradually increase the accuracy of fraud detection.
business online. While there are many benefits to online transactions, like viability, speedier payments, and ease of use, there are drawbacks as well, including fraud, phishing, and data theft. The increasing volume of internet transactions raises the possibility of fraud and dishonest business practices that compromise an individual's privacy. When a criminal can take control of an account and move money out of a person's online bank account. To minimize anticipated financial losses, it is also necessary to enhance traditional machine learning techniques. A feature-engineered machine learning-based model that is Random Forest and Gradient Boosting algorithm which can improve its performance, reinforce its stability, and gain experience by processing as much as data it can. Finally, understanding the costs and risks associated with payment methods is essential to fighting fraud in a methodical and costeffective manner. We define three models to address these issues: a risk model to forecast fraud risk while taking counter measures into account; machine learning-based fraud detection; and economic optimization of machine learning outcomes. Real data is used to test the models.
To progressively improve the accuracy of fraud detection, the automated system that develops and improves the algorithm considers recommendations made by the detectives. With the help of several technologies, we are trying to develop a Web application for machine learningbased fraud detection in this project. An online transaction fraud detection system is necessary to safeguard digital financial transactions and thwart unlawful or fraudulent activities. As more financial transactions are being done online, robust fraud detection techniques are becoming more and more important. The primary goal of these systems is to immediately identify and thwart fraudulent conduct in order to safeguard the integrity and security of online transactions. An online transaction fraud detection system is a set of instruments, algorithms, and processes designed to identify and prevent.
Key Words: Machine Learning, Random Forest, Gradient Boosting, Reinforce, Fraud Detection
1.INTRODUCTION The world is heading quickly toward a cashless society. The number of people making purchases online has increased, according to numerous polls and studies, and it is expected that this trend will continue in the years to come. While this might sound like good news, there is also an increase in fraudulent transactions on the other hand. Despite the deployment of multiple security measures, fraudulent transactions nonetheless result in the loss of a substantial amount of money. When someone uses another person's credit card for unauthorized personal purchases online without the cardholder's or the card issuer's knowledge, this is known as online fraud.
1.1 OBJECTIVE
The process of monitoring user activity to assess, spot, or stop unwanted behaviour, such as fraud, intrusion, and defaults, is known as fraud detection. A person who has fallen for one of these scams frequently does not recognize it until it is too late. Real-world examples show how fast automated systems evaluate the enormous amount of
1.2 CHALLENGES
© 2024, IRJET
|
Impact Factor value: 8.226
To achieve high accuracy in fraud detection while minimizing false positives. Analyze user behavior and transaction patterns. Continuously adapt to evolving fraud tactics and patterns. Seamlessly integrate with various online payment gateways and financial institutions. Enhance user authentication methods. Ensure scalability to handle increasing transaction volumes Identify and mitigate potential risks associated with online transactions, to industry regulations and compliance standards which Build and maintain trust among users.
First off, even while transaction data contains extremely few fraudulent events, the data's imbalance may skew model performance in favour of the majority class, leading to imprecise fraud detection. Second, because fraudsters use dynamic techniques, idea drift occurs, and model adaption is
|
ISO 9001:2008 Certified Journal
|
Page 2499