← Back to directory

Xeek / FORCE 2020 Machine Learning Dataset

Access here

Key facts

Region Norwegian Continental Shelf
Provider Xeek (released for the FORCE 2020 Conference)
Access Download
URL https://xeek.ai/challenges/force-well-logs
Registration No registration
Licence Open for research and educational use (competition licence)
Data types Well / Borehole Metadata, Well Log Data
Coverage A small number of anonymised wells from the Norwegian Continental Shelf, prepared specifically for a machine learning competition.
Formats Tabular well log data suitable for machine learning workflows (commonly CSV); includes labelled target variables for supervised learning.

Notes

The Xeek / FORCE 2020 dataset was released as part of the FORCE 2020 Machine Learning competition and is widely used as a benchmark dataset for subsurface machine learning workflows.

The dataset consists of a small, curated set of anonymised wells with well log measurements and associated lithology or facies labels, designed explicitly for supervised learning tasks such as classification.

The data is intentionally clean and ML-ready, making it well suited for tutorials, experimentation, and benchmarking. However, it does not reflect the scale, heterogeneity, or data quality challenges typically encountered in regulatory or operator datasets, and should be treated as a controlled benchmark rather than a realistic field archive.