Dataset

Expert Label Noise: 176 Annotators, Known Truth

Contains 176 experts' geosteering interpretations, snapshots, trajectories, and truth curves for measuring expert disagreement and error.

Kaggle ↗Jul 20, 2026Code ↗

  • tabular
  • label-noise
  • expert-annotations
  • geosteering
Well trajectory predictions vs geological corridor showing model output against target zone
Well trajectory predictions vs geological corridor showing model output against target zone

What it is

176 expert interpretations of the same known subsurface let you measure real annotation noise and test label aggregation against ground truth. 176 geosteering specialists' readings across 16 scenarios, with final gridded interpretations, intermediate snapshots, drilled trajectories, and known truth curves in 5 tidy CSVs.

Key results

  • 176 specialists across 16 scenarios
  • 5 CSVs: 88,937 final grid rows, 163,499 snapshot rows
  • CC BY 4.0; 27 MB total

Useful for benchmarking annotator aggregation, consensus-vs-budget curves, and agreement metrics against known truth.

One useful ML email a week.

Worth your time, not all of your time.

We track new ML work across papers, code, and data, then keep only the pieces worth a closer look. Selected, not collected.

Papers.Data.Code

The Weekly ML Signal

Selected, not collected.

One useful ML email a week: the papers, repos and datasets worth your attention.

Worth your time, not all of your time.