FraudTrust Sample, Synthetic Temporal Stress StudyExecutable model · logistic manifest
Dataset
synthetic_reference_window.csv
Windows
300 rows → not yet run
Run state
Composing scenario
OBSERVED UPLOAD

New scenario

Applying a controlled, deterministic transformation to reference window synthetic_reference_window.csv (300 rows). Enable one or more operators below; the same source, operators, and seed will always reproduce the same result.

Available attribution features in this window: velocity_24h_txn_count, velocity_7d_amount_sum, merchant_risk_score, channel_field, amount_zscore, device_trust_score

Capability: executable model stress test. Raw-input operators rerun the attached logistic model; label and legacy evidence-layer operators explicitly keep model outputs fixed.

Operator stack (fixed application order, gate to comparison)

No operators enabled yet, the scenario would run unchanged from the reference window.

Same source + operators + seed always reproduces the same result.