FRAUD DETECTION IN BANKING DATA BY MACHINE LEARNING TECHNIQUES

Abstract

The rapid growth of digital banking, online payments, mobile banking, and electronic transactions has increased the risk of fraudulent activities in the financial sector. Traditional rulebased systems often struggle to identify new and evolving fraud patterns in real time. This project proposes a machine learning-based framework for detecting fraudulent banking transactions using historical transaction data. The system considers features such as transaction amount, transaction type, account activity, customer behaviour, transaction time, and other relevant attributes. Data preprocessing includes handling missing values, removing duplicates, encoding categorical features, scaling numerical attributes, and addressing class imbalance through suitable sampling techniques. Feature selection is applied to identify important characteristics associated with fraudulent behaviour. Logistic Regression, Decision Tree, Random Forest, Support Vector Machine (SVM), and XGBoost classifiers are trained and evaluated. Performance is measured using Accuracy, Precision, Recall, F1-Score, ROC-AUC, and Confusion Matrix. The proposed framework enables efficient fraud identification and can assist financial institutions in reducing financial losses and improving real-time transaction monitoring and security. Keywords: Banking Fraud Detection, Machine Learning, Financial Transactions, Random Forest, SVM, XGBoost, Fraud Classification, Cybersecurity.