Leveraging machine learning to minimize fraudulent transactions: a strategic modeling approach

Abstract

Fraud has increasingly gained prevalence as millions of transactions are done online. Various stakeholders such as governments, organizations, and consumers have developed strategies to detect fraud and other unusual behavior. Machine learning techniques have been leveraged for fraud detection resulting in unique and sustainable solutions in financial transactions. In the modern age, machine learning algorithms have been widely utilized as a data mining technique for identifying issues with transactions. The current research aims to compare the effectiveness of three distinct machine learning models including GaussianNB, XGBoost, and Logistical Regression models by focusing on their precision, recall, and F1 score. Based on the outcomes of the three machine learning models, XGBoost is considered to be the best alternative for fraud detection at WeGoWin.