Anomaly detection · 02

Transaction Fraud Detection

A classification study focused on the challenges of rare-event detection, cost-sensitive evaluation, resampling, and practical fraud alerts.

PythonXGBoostImbalanced-learnMatplotlib

Designed for a simple, confident user experience.

A fraud-risk model designed for highly imbalanced transaction data with probability thresholds and precision-recall evaluation.

  • Imbalanced-data strategy
  • Anomaly pattern analysis
  • Precision-recall optimisation
  • Risk scoring dashboard
Back to portfolio →