For Data Scientists & ML Engineers · Oil Gas Drilling
Oil & Gas Drilling Data for Data Scientists
Oil Gas Drilling data for data scientists: 15 datasets on one shelf. Every one delivered as API, files, or warehouse rows.
financial time series api for backtesting · alternative data for quantitative research · where to get training data for oil gas drilling models
API, files, or your warehouse. Daily, weekly, or hourly.
What can you train machine-learning models on?
DataDRILL - Formation Pressure Prediction and Kick Detection Dataset is the one record here that ships ready-made labels: drilling sensor streams in plain CSV tagged for formation-pressure prediction and kick-detection, so a model trains without a labeling pass. It is distributed as a static archive, so checksum the file rather than expecting revisions.
For tabular and spatial features at national scale, USGS's aggregated history covers every US well from 1817 to 2022 in GeoPackage and Esri File Geodatabase - join well coordinates to your own grid cells and derive vintage, spacing and density features. State portals fill in the engineered-feature layer: Colorado ECMC's daily extracts pair CSV tables with shapefile and geodatabase geometry, and North Dakota publishes XLSX and JSON alongside its FeatureServer.
Format coverage matters at pipeline scale: across all 1,744 datasets Datadory catalogs, CSV is offered by 734 (42.1%) and JSON by 675 (38.7%).
Which APIs feed a notebook directly?
NDIC exposes an ArcGIS FeatureServer over 43,824 wells that reads straight into a GeoDataFrame, backed by daily HTML reports.
How do you load decades of well history into a warehouse?
Texas Railroad Commission ships weekly well, lease and production files for the biggest US basin as fixed-width ASCII and EBCDIC extracts, with CSV, JSON, dBase and shapefile copies and PDF field layouts - write the layout parser once and the whole archive opens. Colorado ECMC posts daily ZIP bundles mixing CSV, shapefile and file geodatabase layers.
Two historical compilations reward batch ingestion. EMODnet Human Activities harmonizes 28,506 offshore wells into GIS form, GeoJSON and shapefile on an annual cycle. Wyoming Statewide Time-Stamped Oil and Gas Activity covers 203,000 wells from 1900 to 2020 with explicit start and stop years - the structure you want for survival analysis or activity-duration features.
Can you backtest rig counts as a financial time series?
Yes, and the two standard signals sit at opposite ends of the price axis. EIA's Crude Oil and Natural Gas Drilling Activity (Rotary Rigs) publishes a monthly rig series reaching back to 1949, pulled through a bulk JSON API as xls, html_table or json_api payloads - deep enough history for regime models.
For context variables, the EIA Drilling Productivity Report keeps archived regional productivity tables in XLSX, and AER's Alberta Energy Outlook (ST98) companion XLSX supports Canadian activity modeling. Permit flows make the natural leading indicator: OilPriceAPI and Colorado ECMC both publish permits ahead of rig mobilization.
Straight answers
Is there a financial time series API for backtesting drilling activity?
Yes. EIA's Crude Oil and Natural Gas Drilling Activity (Rotary Rigs) publishes a monthly rig-count series reaching back to 1949 through a bulk JSON API - enough history for regime-switching and cycle models.
What alternative data suits quantitative research on drilling?
Wyoming's 1900-2020 panel adds start and stop years for survival-style features.
Where can I get training data for kick-detection models?
Start with the DataDRILL Formation Pressure Prediction and Kick Detection Dataset: labeled drilling-sensor time series in plain CSV, built for exactly those two tasks. It ships as a static archive rather than a feed, so download once, checksum it and treat any retrain as a versioned event. BSEE's 57,482 offshore wells supply context features.
Rows before rollout
Sample rows from any shelf entry — the field dictionary and coverage notes ride along. If the shelf misses what you need, say so; sourcing requests are half our job.
Talk to us