Optimizing fraud detection in financial transactions with machine learning and imbalance mitigation
Problem
Financial fraud is rising with digital payments, and fraud detection systems need improved predictive performance.
Why it matters
Fraud in digital payments causes direct financial losses and undermines trust in online commerce.
What the AI did
Six machine learning models with optimized hyperparameters were trained on preprocessed transaction data; XGBoost and Random Forest performed best.
Results
XGBoost and Random Forest outperformed the other models; the abstract truncates before exact metric values.
Impact
Peer-reviewed evidence that ensemble ML improves fraud detection; real-world adoption is not evidenced.
Record
| Venue | Expert Systems |
|---|---|
| Published | 2024 |
| Work type | article |
| Authors | Ezaz Aldahasi, Rama Khaled Alsheikh, Fakhri Alam Khan, Gwanggil Jeon |
| Citations | 66 (OpenAlex, at import time) |
| References | 34 |
| Model or method | XGBoost and Random Forest (best of six models) |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1111/exsy.13682 |
| Open access | closed |
| Keywords | fraud detection, machine learning, XGBoost, class imbalance, digital payments |
Importance rating
significant — thousands to millions affected
Abstract calls digital-payment fraud a critical issue with an alarming rise in cybercrime and increasing threats; industry-wide problem, but no quantified scale stated.