Fraud Detection Model (XGBoost)
Binary classifier trained on the Kaggle Credit Card Fraud Detection dataset
(284,807 transactions, ~0.17% fraud). Handles class imbalance via
scale_pos_weight. Test set performance: 0.84 precision / 0.84 recall on
the fraud class, ROC-AUC 0.978.
Input
JSON with Time, V1-V28, Amount fields.
Output
{"prediction": "fraud" | "legit", "fraud_probability": float}