3w-causal-real-group-audit

3W well-event audit tables (Petrobras 3W 2.0)

Decision-level and episode-level results from a well-event detection study on real Petrobras 3W 2.0 recordings, tested on well groups the detectors never saw in training. Use these tables to see how detectors transfer across wells. This is not a new sensor dataset; raw sensor values stay with the upstream source.

from datasets import load_dataset
decisions = load_dataset("sankalpsthakur/3w-causal-real-group-audit", "decisions")

Configurations

• Scores, labels and provenance columns are for auditing. Never feed them to a predictor as inputs.
• event_timestamp_ns is the dataset's nominal clock, not arrival or decision time.
• Filename groups are anonymized public IDs, not authenticated well identities.

What the tables show

The validation-selected detector covered 24 of 55 confirmation episodes and caught 3 of 12 observable onsets within 300 seconds. It raised 42 false alarms over 9.124 normal days and spent 64.392% of normal time in alarm. Use the tables to study transfer, abstention and alarm burden, not barrier integrity, remaining life or deployment readiness.

The study used 357 training, 33 validation, 138 confirmation and 113 reserved recordings, and excluded 478 overlapping real recordings. Models and the full report are in well-production-event-causal.

Source and attribution

Derived from Petrobras 3W 2.0 under CC BY 4.0. The source descriptor was published in April 2026; the pinned release predates it. See ATTRIBUTION.md, table schemas, protocol and manifest. The previous card is in HISTORY.md.