UPI FRAUD TRANSACTION DETECTION USING ML

Abstract

The rapid growth of digital payment systems and Unified Payments Interface (UPI) transactions has transformed the financial sector by enabling fast, secure, and convenient money transfers. However, the increasing popularity of UPI platforms has also led to a significant rise in fraudulent activities such as phishing, fake payment requests, identity theft, account hacking, and unauthorized transactions. Traditional fraud detection techniques often fail to identify sophisticated and real-time fraud patterns due to the massive volume and velocity of digital transactions. To address these challenges, Machine Learning (ML) techniques provide an intelligent and automated approach for detecting fraudulent UPI transactions with higher accuracy and efficiency. This project presents a Machine Learning-based UPI Fraud Transaction Detection System designed to identify suspicious transaction activities in real time. The system collects transaction-related data such as transaction amount, transaction frequency, device information, location, payment time, and user behavior patterns. Data preprocessing techniques including data cleaning, normalization, and feature extraction are applied to improve the quality of the dataset. Various ML algorithms such as Logistic Regression, Decision Tree, Random Forest, Support Vector Machine (SVM), and XGBoost are used to classify transactions as genuine or fraudulent. The proposed system analyzes transaction patterns and predicts fraudulent behavior based on previously learned data. It helps financial institutions and payment platforms reduce financial losses and improve customer trust. Performance evaluation metrics such as accuracy, precision, recall, and F1-score are used to measure the effectiveness of the models. Experimental results show that ensemble learning techniques like Random Forest and XGBoost provide better fraud detection performance compared to traditional methods.