Payment Fraud Radar

Real-time ML Transaction Risk Analytics

Model Architecture Random Forest
Decision Threshold 0.4694
Validation Recall 99.8%
Validation Precision 100.0%
Model Status Active (2026-09-03)
Notice: This is a machine-learning estimate from a model trained on a public, simulated dataset. It is not a guaranteed financial decision and must not be used on its own to block, approve or investigate a real payment.

Transaction Parameters

Input Data
Quick Test Scenarios:
Channel used for the payment.
Value being moved.
Balance on the sending account before the transaction.
Balance on the sending account after the transaction.
Balance on the receiving account before the transaction.
Balance on the receiving account after the transaction.
0-23. Genuine traffic collapses overnight, fraud does not.

Predictions are calculated using local scikit-learn pipeline.

Assessment Result

Awaiting Input

No Transaction Analyzed Yet

Select a quick scenario or submit custom details to view real-time risk scores and engineered signal notes.

0% Fraud Prob.
Verdict:
Assessed Risk Level: -

Model Decision Factors & Signals