<![CDATA[This study aims to develop a credit card transaction fraud detection model using machine learning approaches, namely Logistic Regression, Random Forest, and Gradient Boosting Classifier. The dataset used is sourced from real credit card transactions with a fraud proportion of 0.17%, which reflects the problem of class imbalance. To overcome this, the Synthetic Minority Over-sampling Technique (SMOTE) and feature transformation using Principal Component Analysis (PCA) were applied. The evaluation was carried out using accuracy, precision, recall, F1-score, and ROC-AUC metrics. The results show that Random Forest and Gradient Boosting Classifier produced the best performance with near-perfect accuracy and ROC-AUC values (ROC-AUC > 0.999), while Logistic Regression gave very good results but slightly below the other two models. However, the near-perfect ROC-AUC value may indicate potential overfitting, requiring further validation on different datasets. Unlike previous studies that only used one algorithm, this study compared three models simultaneously and integrated SMOTE and PCA to improve detection performance. The practical implication of this study is that the proposed model can be implemented in digital financial systems to assist banking institutions in detecting fraud in real time and reducing potential financial losses.]]>
Copyrights © 2025
