captured_on
dateUTC day the prediction was captured.
per day
- Valid
- 3,669 · 100%
- Missing
- 0 · 0%
- Distinct
- 3 · 0.1% of valid
81.1 kB across 1 part, newest sealed 22 November 2026 at 12:42 UTC. The first build wrote 1 part and 3,669 rows.
UTC day the prediction was captured.
per day
EIA code of the balancing authority.
EIA region.
Start of the predicted hour, UTC.
per day
live: captured before EIA published the hour.
Corrected forecast as captured.
10th percentile as captured.
90th percentile as captured.
The operator's own day-ahead forecast at capture.
Pinned model.
Newest measured hour the model had seen.
per day
How far ahead the prediction was: hours after the operator's newest published demand hour on the day it was captured (its first predicted hour is 1). Counted per capture day.
Demand EIA published; empty until it is published.
Corrected forecast minus actual, percent of actual.
Operator forecast minus actual, percent of actual.
Whether the actual fell inside the captured p10 to p90 band.
Whether the corrected forecast was closer to the actual than the operator's.
Profile of an earlier version, all rows, computed 22 Nov 2026, 07:13 UTC
| Passed | Freshness | Latest event 18 min after that version |
|---|---|---|
| Warning | Gaps | 48 empty hours, 18 Nov 05:00 to 20 Nov 04:00 |
| Warning | Anomalies | Gap on 18 Nov 05:00, 0 rows against about 1, and 6 more |
| Passed | Unique keys | Every captured_on · balancing_authority · hour_utc is unique |
| Passed | Change since last profile | No change, 3,669 rows |