ONLINE FRAUD PAYMENT DETECTION USING BALANCED ML ALGORITHM

Abstract

Credit card fraud has become a major concern in digital transactions, leading to significant financial losses for individuals and organizations. This project focuses on developing a machine learning-based system to detect fraudulent transactions using a highly imbalanced dataset containing approximately 28,000 records. The dataset consists of two classes: fraud and non-fraud transactions, where non-fraud cases significantly outnumber fraudulent ones. To address this imbalance, the system applies the Synthetic Minority Oversampling Technique (SMOTE) to generate synthetic samples and balance the dataset, improving model performance. The system implements multiple machine learning algorithms such as Decision Tree, Random Forest, and other classifiers to detect fraud. Models are trained on both original and balanced datasets to compare performance. Experimental results show that applying SMOTE significantly improves detection accuracy and reduces bias toward the majority class. Among the models, Random Forest with SMOTE achieves the highest accuracy, exceeding 99%, demonstrating its effectiveness in handling imbalanced data. The system also uses confusion matrix and graphical visualizations to evaluate model performance and highlight correct and incorrect predictions. A user-friendly interface allows users to load datasets, balance data, train models, and test new transaction data for fraud detection. The system predicts whether a transaction is fraudulent or normal in real time. Overall, this project demonstrates the importance of handling imbalanced data and the effectiveness of machine learning techniques in fraud detection.