<![CDATA[Savings and loan cooperatives play a crucial role in supporting the community's economy. With the motto "From Members, By Members, and For Members," cooperatives focus on developing member funds and providing returns in the form of dividends. However, cooperative operations are not free from the risk of fraud, especially by internal parties (employees or administrators). This study aims to develop a machine learning-based transaction fraud detection model using the Extreme Gradient Boosting (XGBoost) algorithm and to increase model transparency through an Explainable Artificial Intelligence (XAI) approach with the SHAP (SHapley Additive exPlanations) method. This study uses user activity log data and financial transactions that can be described as operator/employee behavior in the savings and loan cooperative system. The model will be trained to classify whether transactions are fraudulent or non-fraudulent. The results will then be evaluated using accuracy, precision, recall, F1-score, and AUC metrics. The results show that the XGBoost model has good performance with an accuracy value of 0.81 and an AUC of 0.912. SHAP analysis shows that features such as transaction amount, transaction frequency, transaction time, and changes in user and member data are key factors in fraud detection. This study demonstrates that the integration of XGBoost and SHAP can improve fraud detection accuracy and provide transparency in model decision-making. Therefore, the results of this study can support a more effective supervisory system for savings and loan cooperative financial institutions.]]>
Copyrights © 2026
