model_name: digital-banking-fraud-detector
license: gpl-3.0
language: en
tags:
- tabular
- classification
- fraud-detection
- financial
- synthetic-data
metrics:
- accuracy
- f1
- roc_auc
Digital Banking Fraud Detector
Model Overview
Digital Banking Fraud Detector is a machine learning model trained to detect fraudulent financial transactions.
It uses synthetic banking transaction data and is intended for research, tutorials, and ML demos.
Model Details
- Model Name: digital-banking-fraud-detector
- Model Type: Binary Classification
- Framework: scikit-learn
- Input Format: CSV / Tabular data
- Output: Fraud label (Yes / No)
Training Data
Synthetic dataset with the following features:
| Feature | Type | Description |
|---|
| transaction_id | String | Unique transaction ID |
| customer_id | String | Customer ID |
| transaction_type | Categorical | Transfer, Payment, Withdrawal, Deposit |
| channel | Categorical | Mobile App, Web, ATM, Branch |
| amount_usd | Float | Transaction amount |
| currency | Categorical | USD |
| country | Categorical | Country of transaction |
| merchant_category | Categorical | Category of merchant |
| is_fraud | Binary | Target label |
Intended Use
✅ Fraud detection demos
✅ ML experimentation
❌ Real financial decision-making
Evaluation Results
| Metric | Score |
|---|
| Accuracy | 0.92 |
| F1-score | 0.89 |
| ROC-AUC | 0.95 |
Limitations
- Synthetic data only
- Limited geographic coverage
- Simplified transaction patterns
Ethical Considerations
This model does not use real customer data.
Misuse in production could unfairly block legitimate transactions.
License
This model is released under the GPL-3.0 License.
