Calibration
Loading public event evidence
The validation register is opening the read-only event view.
Source files are never used as a runtime fallback.
Loading public event evidence
Source files are never used as a runtime fallback.
MODEL VALIDATION / ANALYST TRIAGE DEMONSTRATION
Full row-level event queries, fixed holdout evidence, and constrained review capacity.
Measured 1% analyst-review queue only. Not production-ready.
CONTROL 01 / PUBLIC EVENT QUERY
Query every allowlisted event row. Results are read-only, ordered by event time, and limited to 100 rows per request.
| Event time, UTC | Event ID | Merchant / category | Amount | Fraud label | Source |
|---|---|---|---|---|---|
| Jan 1, 2019, 7:30 PM | 1876 | fraud_Mosciski Grouptravel | $6.72 | NO | fraudTrain |
| Jan 1, 2019, 7:30 PM | 1877 | fraud_Rau-Grantkids_pets | $154.47 | NO | fraudTrain |
| Jan 1, 2019, 7:31 PM | 1878 | fraud_Jakubowski Inchome | $109.41 | NO | fraudTrain |
| Jan 1, 2019, 7:31 PM | 1879 | fraud_Greenholt Ltdhealth_fitness | $73.81 | NO | fraudTrain |
| Jan 1, 2019, 7:31 PM | 1880 | fraud_Baumbach Ltdpersonal_care | $71.86 | NO | fraudTrain |
| Jan 1, 2019, 7:32 PM | 1881 | fraud_Durgan, Gislason and Spencerhome | $20.60 | NO | fraudTrain |
| Jan 1, 2019, 7:32 PM | 1882 |
CONTROL 02 / CAPACITY
Change the planning rate to see workload. Recall remains visible only for the measured 1% policy because no other threshold sweep is available.
EVIDENCE 03 / MODEL COMPARISON
Both policies train on the earlier simulated partition and evaluate on the later one. The challenger only adds class weighting to the same logistic model family. It loses on every reported measure and is not promoted.
Calibration
CONTROL 04 / DATA BOUNDARY
The complete simulated source is published through allowlisted analytical fields. Identity-like raw columns and decisioning capabilities remain outside the public view.
Not available here: identity-like raw columns, personal identifiers, model scores, scoring endpoints, payment decisions, or payment-processing behavior.
EVIDENCE 05 / COVERAGE LEDGER
These source totals reconcile the public event view to all 1,852,394 ingested rows.
| $40.83 |
| NO |
| fraudTrain |
| Jan 1, 2019, 7:32 PM | 1883 | fraud_Kuphal-Toyfood_dining | $59.59 | NO | fraudTrain |
| Jan 1, 2019, 7:32 PM | 1884 | fraud_Erdman-Durganhealth_fitness | $168.64 | NO | fraudTrain |
| Jan 1, 2019, 7:32 PM | 1885 | fraud_Homenick LLCpersonal_care | $43.76 | NO | fraudTrain |
| Jan 1, 2019, 7:33 PM | 1886 | fraud_Metz-Boehmshopping_pos | $2.17 | NO | fraudTrain |
| Jan 1, 2019, 7:34 PM | 1887 | fraud_Reichert, Huels and Hoppeshopping_net | $52.57 | NO | fraudTrain |
| Jan 1, 2019, 7:35 PM | 1888 | fraud_O'Keefe-Wisokyfood_dining | $1.29 | NO | fraudTrain |
| Jan 1, 2019, 7:35 PM | 1889 | fraud_Runolfsdottir, Mueller and Handentertainment | $46.28 | NO | fraudTrain |
| Jan 1, 2019, 7:35 PM | 1890 | fraud_Fisher-Schowaltershopping_net | $1.30 | NO | fraudTrain |
| Jan 1, 2019, 7:35 PM | 1891 | fraud_Torp, Muller and Borerhealth_fitness | $29.48 | NO | fraudTrain |
| Jan 1, 2019, 7:36 PM | 1892 | fraud_Nienow PLCentertainment | $20.76 | NO | fraudTrain |
| Jan 1, 2019, 7:36 PM | 1893 | fraud_Runte-Mohrfood_dining | $55.57 | NO | fraudTrain |
| Jan 1, 2019, 7:37 PM | 1894 | fraud_Morissette PLCshopping_pos | $8.31 | NO | fraudTrain |
| Jan 1, 2019, 7:38 PM | 1895 | fraud_Beier and Sonshome | $49.93 | NO | fraudTrain |
| Jan 1, 2019, 7:38 PM | 1896 | fraud_Jerde-Hermanntravel | $1.73 | NO | fraudTrain |
| Jan 1, 2019, 7:39 PM | 1897 | fraud_Botsford and Sonshome | $262.12 | NO | fraudTrain |
| Jan 1, 2019, 7:39 PM | 1898 | fraud_Denesik, Powlowski and Pouroshome | $37.15 | NO | fraudTrain |
| Jan 1, 2019, 7:39 PM | 1899 | fraud_Kihn, Brakus and Goyettepersonal_care | $3.09 | NO | fraudTrain |
| Jan 1, 2019, 7:42 PM | 1900 | fraud_Metz-Boehmshopping_pos | $6.53 | NO | fraudTrain |
The dashed diagonal is perfect calibration. The observed curve is uneven, so scores should not be interpreted as reliable probabilities without further calibration work.
Review capacity
Review capacity stays fixed. Monthly volume changes with the holdout distribution. This is retrospective workload evidence, not a forecast.
Limits: simulated source; one chronological split; allowlisted amount, merchant, category, and prior-history inputs; no confidence intervals, fairness analysis, threshold sweep, live scoring, or evidence for automated payment decisions.