Development of a Predictive Model for Fraud Detection in Online Transactions Using Machine Learning Algorithms

The rise in fraudulent activities within banking operations has brought to light a significant problem. Conventional fraud detection methods often lack accuracy and fail to keep up with the rapidly evolving fraud techniques. This research focuses on developing and enhancing fraud detection methods using machine-learning techniques to overcome the unresolved challenges in current techniques, address imbalanced datasets, improve model resilience against adversarial attacks, and enhance real-time detection performance. The proposed predictive model employed optimized machine learning algorithms (logistic regression, decision tree, K-nearest neighbors (KNN), Support Vector Classifier (SVC), Naive Bayes, and Random Forest) in building a system to achieve real-time fraud detection, providing timely alerts and minimizing potential financial losses. The hyperparameter tuning process involved utilizing GridSearchCV to search through a predefined parameter grid and identify the optimal combination of hyperparameters. Subsequently, the model with the optimized hyperparameters was evaluated on the test data. The test accuracy of the tuned Random Forest classifier was found to be 0.9907, indicating a high level of accuracy in identifying fraudulent transactions. The important part of this work is the ability of the model to predict fraudulent transactions with high degrees of accuracy.