Smart Card Fraud Detection using Ensemble Methods in Machine Learning

This study examines the growing use of cards, online payments and the associated risk of fraud, which has led to significant financial losses globally. Financial institutions like banks need to detect fraudulent transactions effectively, prompting an analysis of different machine learning classifiers. The findings reveal that the Decision Tree algorithm outperformed Random Forest, achieving higher accuracy and superior fraud detection rates. Random Forest, on the other hand, struggled to handle the highly imbalance in the dataset, leading to notably lower accuracy. To overcome these limitations and enhance fraud detection capabilities, a hybrid model was developed, combining the strengths of both Decision Tree and Random Forest. This ensemble method achieved enhanced accuracy and robustness by leveraging the interpretability of Decision Trees and the feature-selection capabilities of Random Forest. The paper also highlights the importance of using machine learning to prevent fraudulent transactions and improve transaction security.