MACHINE LEARNING FOR DETECTING CYBERCRIME IN THE BANKING SECTOR
Ghofran Anis, Ghofran Anis · Journal of Southwest Jiaotong University · 2023
In the digital age, the escalating threat of cybercrime poses a significant concern, impacting critical areas such as privacy, national security, societal norms, and intellectual property rights. This research paper addresses the urgent challenge of cybercrime detection within the banking sector, with particular emphasis on safeguarding customer information and preventing financial losses. In this study, we employed various classification models to forecast instances of cybercrime in the banking industry. Specifically, we explored and compared the performance of 11 machine learning models, including K-nearest neighbors (KNN), random forest (RF), naive Bayes, gradient boosting, multi-layer perceptron (MLP), decision tree, AdaBoost, support vector machine (SVM), linear support vector classification (linear SVC), a voting classifier, and XGBoost algorithms for binary classification within the banking domain. The primary objective of this research is to enhance the accuracy and efficacy of cybercrime detection within the banking sector. We used a comprehensive dataset with features such as login attempts, transaction amounts, device attributes, and geographic location. The evaluation of model performance included critical metrics such as accuracy, precision, recall, and F1 score. The study’s outcomes highlighted the exceptional performance of several models, with the RF model emerging as the top performer, achieving a remarkable accuracy rate of 99.99%. In addition, other models, including gradient boosting, decision tree, AdaBoost, SVM, and XGBoost, closely followed with an accuracy of 99.98%. These findings underscore the remarkable ability of these models to accurately identify instances of cybercrime, offering promising prospects for enhanced cybersecurity within the banking sector. Keywords: Cybercrime, Machine Learning, Banking Sector, E-Banking DOI： https://doi.org/10.35741/issn.0258-2724.58.5.60