UPI Fraud Detection Using ML

Abstract—The Unified Payments Interface (UPI) has drastically changed how digital transactions are conducted in India. But along with this expansion comes a rise in fraudulent activity that takes advantage of holes in digital infrastructure. A machine learning-based fraud detection system designed specifically for UPI transactions is shown in this study. The system detects abnormal behavior with high accuracy by using Random Forest and K-Nearest Neighbors (KNN) algorithms on a carefully selected dataset of transactional information (Transaction ID, Timestamp, Amount, Sender/Receiver UPI IDs, and Status). To deal with dimensionality and class imbalance, methods like data resampling and Principal Component Analysis (PCA) are used. The system's ability to boost security in digital payment ecosystems is demonstrated by experimental results that show better fraud detection performance. Keywords—PCA, Random Forest, KNN, Machine Learning, Fraud Detection, UPI, and Class Imbalance