MACHINE LEARNING MODELS FOR FINANCIAL FRAUD DETECTION IN E-COMMERCE PLATFORMS

Abstract

The rapid digitization of financial transactions in e-commerce platforms has increased the risk of fraudulent activities, prompting the need for robust fraud detection mechanisms. Machine learning (ML) offers dynamic, data-driven solutions that can adapt to evolving fraud patterns in real-time. This study investigates the application of supervised and unsupervised ML models for financial fraud detection in e-commerce, evaluates their performance, and explores hybrid approaches for enhanced accuracy. Utilizing datasets from simulated e-commerce transactions, the study compares algorithms such as Logistic Regression, Decision Trees, Random Forest, Support Vector Machines (SVM), and neural networks. Findings highlight the effectiveness of ensemble and deep learning methods in detecting complex fraudulent behavior while maintaining low false-positive rates. The research concludes with recommendations for integrating ML models into e-commerce platforms for proactive fraud prevention.