Open access Jurnal SINTA 2
Penerapan Metode Stacking Ensemble Learning untuk Deteksi Penipuan pada Transaksi Finansial
Implementation of Stacking Ensemble Learning Method for Fraud Detection in Financial Transactions
Jurnal Teknologi Informasi dan Ilmu Komputer (JTIIK) · Vol. 13 No. 3 (2026) · pp. 521–530
- ISSN
- 2528-6579, 2355-7699
- Published
- 2026-06-30
- Publisher
- Fakultas Ilmu Komputer, Universitas Brawijaya
- Language
- id
Abstract
Bahasa Indonesia
Identifikasi fraud pada transaksi keuangan menjadi pekerjaan besar untuk mencegah kerugian yang dapat terjadi akibat fraud. Dengan perkembangan teknologi machine learning yang makin baik, eksperimen mengenai identifikasi fraud dapat dilakukan dengan menggunakan model machine learning. Sejumlah model machine learning dapat digunakan untuk identifikasi fraud, seperti logistic regression, XGBoost, LightGBM, dan Stacking. Penelitian ini untuk menunjukkan kinerja dari stacking ensemble yang dipadukan dengan MLP sebagai meta learner. Identifikasi fraud memiliki tantangan tersendiri berupa banyak data yang tidak seimbang antara transaksi otentik dengan transaksi palsu. Meskipun demikian, eksperimen yang dilakukan menggunakan model dan evaluasi yang tahan terhadap data yang tidak seimbang. Eksperimen dilakukan dengan pengolahan data dan feature engineering untuk mengatasi permasalahan dalam dataset. Pengolahan data dan feature engineering ditujukan untuk mengurangi pencilan, fitur yang tak relevan, dan nilai-nilai pada data yang hilang atau terduplikasi. Kemudian, pembuatan model dilakukan dengan logistic regression, XGBoost, LightGBM, dan stacking ensemble dengan base learner dari XGBoost dan LightGBM serta meta-learner menggunakan MLP. Hasil dari eksperimen yang dilakukan menunjukkan bahwa stacking memiliki kinerja yang paling tinggi dengan nilai AUC-ROC mencapai 0,9663 dibandingkan model lain.
English
Fraud detection in financial transactions is a critical task to prevent potential losses caused by fraudulent activities. With the advancement of machine learning technology, experiments on fraud identification can now be conducted using various machine learning models. Several models such as Logistic Regression, XGBoost, LightGBM, and Stacking can be employed for this purpose. This study aims to demonstrate the performance of a stacking ensemble model combined with a Multi-Layer Perceptron (MLP) as the meta-learner. Fraud detection poses a unique challenge due to the highly imbalanced nature of the data, where genuine transactions significantly outnumber fraudulent ones. Despite this, the experiments conducted in this study utilized models and evaluation metrics that are robust to imbalanced data. The process involved data preprocessing and feature engineering to address issues within the dataset. These steps were undertaken to reduce outliers, eliminate irrelevant features, and handle missing or duplicated values. Model development was carried out using Logistic Regression, XGBoost, LightGBM, and a stacking ensemble with XGBoost and LightGBM as base learners, and MLP as the meta-learner. The experimental results indicate that the stacking model achieved the best performance, with an AUC-ROC score of 0.9663, outperforming the other models.
Keywords
Collaboration network of the authors
23 authors · 37 co-authorship links · 6 across different organisations
Hover or tap a name to highlight its co-authors. Double-click to open the author's page.
- Bandung Institute of Technology
- Telkom University
- Badan Pusat Statistik
- Cohere (Canada)
- Affiliation unknown
- Different organisations
- Same organisation
- Affiliation unknown
- Thicker line = more joint articles · bigger circle = more articles
Show as table
| Co-author | Organisation | Joint articles |
|---|---|---|
| Dwi Hendratmo Widyantoro | Bandung Institute of Technology | 2 |
| Aswin Juari | Bandung Institute of Technology | 1 |
| Novi Yusliani | Bandung Institute of Technology | 1 |
| Hari Bagus Firdaud | Affiliation unknown | 1 |
| Zakiy Firdaus Alfikri | Bandung Institute of Technology | 1 |
| Untari Novia Wisesty | Telkom University | 1 |
| Tati Rajab Mengko | Bandung Institute of Technology | 1 |
| Adi Pancoro | Bandung Institute of Technology | 1 |
| Afrida Helen | Bandung Institute of Technology | 1 |
| Nila Nurmala | Badan Pusat Statistik | 1 |
| Fariska Zakhralativa Ruskanda | Bandung Institute of Technology | 1 |
| Muhammad Faris Muzakki | Bandung Institute of Technology | 1 |
| Arief Purnama Muharram | Bandung Institute of Technology | 1 |
| Ubaidillah Ariq Prathama | Bandung Institute of Technology | 1 |
| Samuel Cahyawijaya | Cohere (Canada) | 1 |
| Rinaldi Munir | Bandung Institute of Technology | 1 |
| Bagus Setyawan Wijaya | Affiliation unknown | 1 |
| Bagus Setyawan Wijaya | Affiliation unknown | 1 |
Affiliations come from OpenAlex; journal OAI-PMH metadata carries none. Authors without an OpenAlex ID are matched by name, which can merge or split people.