Advanced Fraud Detection: Leveraging K-SMOTEENN and Stacking Ensemble to Tackle Data Imbalance and Extract Insights
This study proposes an innovative solution for credit card fraud detection, utilizing a stacking ensemble of machine learning classifiers enhanced with sophisticated data resampling techniques. The model demonstrates exceptional performance, achieving an F1-score of 0.92, precision of 0.95, recall o...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10836682/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832592958008852480 |
|---|---|
| author | Nurafni Damanik Chuan-Ming Liu |
| author_facet | Nurafni Damanik Chuan-Ming Liu |
| author_sort | Nurafni Damanik |
| collection | DOAJ |
| description | This study proposes an innovative solution for credit card fraud detection, utilizing a stacking ensemble of machine learning classifiers enhanced with sophisticated data resampling techniques. The model demonstrates exceptional performance, achieving an F1-score of 0.92, precision of 0.95, recall of 0.88, an AUPRC of 0.96, and a perfect ROC-AUC of 1.00. These results significantly outperform standalone models like XGBoost and Decision Tree, showcasing the strength of the proposed approach. The study addresses two critical challenges, class imbalance and overfitting, by employing K-means SMOTEENN to balance minority class representation while reducing synthetic data noise, thus lowering the risk of overfitting. Additionally, the stacking ensemble’s integration of diverse classifiers produces a generalized decision boundary, enhancing model robustness in real-world scenarios. This ensures a precise fraud detection mechanism without compromising sensitivity. Furthermore, the proposed model incorporates Explainable AI (XAI) techniques to enhance interpretability and trust. The study identifies key features driving model predictions by leveraging Local Interpretable Model-Agnostic Explanations (LIME), illustrating how base learners and the meta-learner contribute to the final decision. This transparency bolsters stakeholder confidence and provides actionable insights for financial institutions. Overall, the proposed approach marks a notable step in protecting financial transactions from fraud. |
| format | Article |
| id | doaj-art-c7b3edf5d979428fae7b4e8c51c6d981 |
| institution | Kabale University |
| issn | 2169-3536 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj-art-c7b3edf5d979428fae7b4e8c51c6d9812025-01-21T00:01:17ZengIEEEIEEE Access2169-35362025-01-0113103561037010.1109/ACCESS.2025.352807910836682Advanced Fraud Detection: Leveraging K-SMOTEENN and Stacking Ensemble to Tackle Data Imbalance and Extract InsightsNurafni Damanik0https://orcid.org/0000-0001-8157-3940Chuan-Ming Liu1https://orcid.org/0000-0001-9005-5715College of Electrical Engineering and Computer Science, National Taipei University of Technology, Taipei, TaiwanDepartment of Computer Science and Information Science, National Taipei University of Technology, Taipei, TaiwanThis study proposes an innovative solution for credit card fraud detection, utilizing a stacking ensemble of machine learning classifiers enhanced with sophisticated data resampling techniques. The model demonstrates exceptional performance, achieving an F1-score of 0.92, precision of 0.95, recall of 0.88, an AUPRC of 0.96, and a perfect ROC-AUC of 1.00. These results significantly outperform standalone models like XGBoost and Decision Tree, showcasing the strength of the proposed approach. The study addresses two critical challenges, class imbalance and overfitting, by employing K-means SMOTEENN to balance minority class representation while reducing synthetic data noise, thus lowering the risk of overfitting. Additionally, the stacking ensemble’s integration of diverse classifiers produces a generalized decision boundary, enhancing model robustness in real-world scenarios. This ensures a precise fraud detection mechanism without compromising sensitivity. Furthermore, the proposed model incorporates Explainable AI (XAI) techniques to enhance interpretability and trust. The study identifies key features driving model predictions by leveraging Local Interpretable Model-Agnostic Explanations (LIME), illustrating how base learners and the meta-learner contribute to the final decision. This transparency bolsters stakeholder confidence and provides actionable insights for financial institutions. Overall, the proposed approach marks a notable step in protecting financial transactions from fraud.https://ieeexplore.ieee.org/document/10836682/K-MEANSSMOTEENNcredit cardfraud detectionstacking ensembleimbalance technique |
| spellingShingle | Nurafni Damanik Chuan-Ming Liu Advanced Fraud Detection: Leveraging K-SMOTEENN and Stacking Ensemble to Tackle Data Imbalance and Extract Insights IEEE Access K-MEANS SMOTEENN credit card fraud detection stacking ensemble imbalance technique |
| title | Advanced Fraud Detection: Leveraging K-SMOTEENN and Stacking Ensemble to Tackle Data Imbalance and Extract Insights |
| title_full | Advanced Fraud Detection: Leveraging K-SMOTEENN and Stacking Ensemble to Tackle Data Imbalance and Extract Insights |
| title_fullStr | Advanced Fraud Detection: Leveraging K-SMOTEENN and Stacking Ensemble to Tackle Data Imbalance and Extract Insights |
| title_full_unstemmed | Advanced Fraud Detection: Leveraging K-SMOTEENN and Stacking Ensemble to Tackle Data Imbalance and Extract Insights |
| title_short | Advanced Fraud Detection: Leveraging K-SMOTEENN and Stacking Ensemble to Tackle Data Imbalance and Extract Insights |
| title_sort | advanced fraud detection leveraging k smoteenn and stacking ensemble to tackle data imbalance and extract insights |
| topic | K-MEANS SMOTEENN credit card fraud detection stacking ensemble imbalance technique |
| url | https://ieeexplore.ieee.org/document/10836682/ |
| work_keys_str_mv | AT nurafnidamanik advancedfrauddetectionleveragingksmoteennandstackingensembletotackledataimbalanceandextractinsights AT chuanmingliu advancedfrauddetectionleveragingksmoteennandstackingensembletotackledataimbalanceandextractinsights |