Fraud Detection Pipeline
by Community · unknown · Last verified 2026-03-17
This is a complete machine learning pipeline for detecting fraudulent transactions in real-time. It employs a hybrid approach, using XGBoost or LightGBM for classification and an Isolation Forest for anomaly detection. The system is specifically designed to handle severely imbalanced datasets through SMOTE-Tomek resampling and cost-sensitive learning.
https://github.com/scikit-learn-contrib/imbalanced-learn ↗B
B—Above Average
Adoption: B+Quality: AFreshness: ACitations: BEngagement: F
Specifications
- License
- MIT
- Pricing
- unknown
- Capabilities
- Real-time transaction scoring via low-latency API (<10ms), Hybrid modeling with XGBoost/LightGBM and Isolation Forest, Advanced class imbalance handling with SMOTE-Tomek resampling, Cost-sensitive learning to prioritize minority class detection, Model explainability using SHAP for dispute resolution, Automated feature drift monitoring to detect data changes, End-to-end training and deployment script
- Integrations
- Use Cases
- [object Object], [object Object], [object Object], [object Object], [object Object]
- API Available
- Yes
- Language
- python
- Dependencies
- xgboost, lightgbm, imbalanced-learn, shap, fastapi, redis, pandas
- Environment
- Python 3.10+
- Est. Runtime
- Training: 10-30 min; inference: <10ms
- Tags
- fraud-detection, anomaly-detection, imbalanced-learning, xgboost, lightgbm, real-time-scoring, smote, isolation-forest, model-explainability, shap, ml-pipeline, fintech
- Added
- 2026-03-17
- Completeness
- 0.8%
Index Score
63.7Adoption
75
Quality
87
Freshness
85
Citations
65
Engagement
0