Historical screening scores for Colorado wells — not validated future-well probabilities
loading…
loading model status…
⌘ Terminal · Workbench · 3D · Report →
Screens an undrilled point against the play's trained ensemble — same features, same calibration as the wells above. Texas plays only.
762 pending CO permits scored by the strict pre-drill contract. This is a screening queue with stated uncertainty, not a prediction that these wells succeed: Colorado's forward test is weak (rolling-origin mean AUC 0.6218 vs 0.6751 for a point-in-time kNN baseline, only 1 of 5 windows positive). The labelled CO base rate is 0.8605 and the supported inter-quartile spread is just 0.040, so the top of the board is not meaningfully separable from the middle. 41 low-support permits render hollow grey and carry no rank.
USGS assessment units — published geologic extent, not a model output.
Co-located recorded earthquakes (USGS ComCat + TexNet, 5-yr window). Not attributed to the wells scored on this map.
Colorado analyst — it reads the CO · DJ dataset only, so a question that moves the map switches it back to Colorado.