Every project
Real-Time Fraud Detection
A supervised + unsupervised ensemble — catches known patterns and never-seen-before ones.
What it does
Fraud data is extremely imbalanced and attackers keep changing tactics. XGBoost learns the known fraud patterns while Isolation Forest catches outliers that appear in no label set, with a rule-based alert layer combining both.