Supplementary data and code for "Reliability Boundaries, Drift Diagnosis, and Local Recalibration of Well-Log Machine-Learning Predictions: A Case Study of TOC and Total Gas Content in the Ordovician Wulalike Formation, Western Ordos Basin, China"

This package accompanies the manuscript "Reliability Boundaries, Drift Diagnosis, and Local Recalibration of Well-Log Machine-Learning Predictions: A Case Study of TOC and Total Gas Content in the Ordovician Wulalike Formation, Western Ordos Basin, China" submitted to Natural Resources Research. It contains the machine-readable supplementary Tables S1-S25 (CSV) and the analysis code (scripts, environment specification, fixed random seeds, fold definitions, out-of-fold predictions, and a smoke test) needed to regenerate the reported metrics and tables. The raw well-log, core, and laboratory datasets are confidential oilfield data and are available from the corresponding author on reasonable request; well identifiers in the released files are anonymized.