Fast Fraud Screening Using Lightweight Models - Built a two-stage fraud detection pipeline that reduces computational costs while maintaining high accuracy. The lightweight screening model flags suspicious transactions for deeper analysis, significantly improving processing speed without sacrificing detection quality. Erdos Institute Group Project
data-science machine-learning data-visualization xgboost hyperparameter-optimization hyperparameter-tuning fraud-detection scikitlearn-machine-learning netwrokx
-
Updated
Dec 15, 2025 - Jupyter Notebook