Real-time inference

Credit card fraud risk scoring

Submit a transaction manually or upload a batch as CSV. Each record is scored by a gradient-boosted ensemble, and returned with a calibrated risk level and an approximate feature-influence breakdown.

Scored this session

Flagged as fraud

Avg. probability

—

Avg. latency

—

Transaction input

Manual

Batch CSV

AMOUNT ($)

TIME (sec)

PCA components (V1–V28) · highlighted fields carry the most model weight

Model: XGBoost · threshold auto-applied server-side

Risk assessment

—

fraud probability

AWAITING INPUT

Threshold used—

Risk level—

Latency—

Session probability trend

Run a prediction to start plotting the trend.

Top feature influence

Run a prediction to see which inputs drove it.

Session history

No transactions scored yet this session.