dc_002 · Deployment: cold start (first-ever fire)
Status: ✓ Passing · Last run: 2026-06-09T06:20:30 · Pipeline: deployment
· History: 1/1 runs passed (100%)
What was tested
Deployment is brand new — no prior_predictions table exists. First scheduled fire should score the new rows and produce an output table cleanly, without choking on the missing history.
What we planted in the data
- 1,000 entities
- Signal strength: 0.50 (sigmoid slope multiplier)
n_new_rows=100
What the system did
- Training rows built: 1,000
- Features used in model: 3
- Model selected:
DeploymentScoringRunner
What we checked — all assertions passed
| Status | Assertion | Detail |
|---|---|---|
| ✓ | deployment_completed_cleanly |
clean run |
| ✓ | scored_row_count_eq |
scored 100 rows, expected 100 |
| ✓ | prediction_distribution_in_band |
predictions in [0.002, 1.000] (mean 0.488) |
| ✓ | prior_predictions_unchanged |
no prior predictions (cold-start) — idempotency check vacuous |
How this could have gone wrong
(no assertion descriptions provided)
Why this case matters
Every deployment starts as a cold start. If the first fire fails, the deployment never gets off the ground.
Reproducing
# from auto_insight_api/
python -m validation.v2 run dc_002 --pipeline deployment -v
- Case config:
validation/v2/cases/dc_002_cold_start.yaml - Data shape:
deployment_cold_start - Analytics type:
deployment