Supporting dataset for Triggering Larger Earthquakes on Moderately Unstable Faults under Long-Term Injection: Insights from The Geysers, California

This dataset contains the processed earthquake features used for the machine learning analysis in Wang et al. (2025), submitted to Journal of Geophysical Research: Solid Earth. The dataset (features_for_ml.csv) includes the following columns: ID: event identifier Mw: earthquake magnitude Inst I: instability index V_weighted: distance-weighted injection volume (million m³) Dist_to_9: distance to well Prati 9 (m) Dist_to_29ab: distance to wells Prati 29a/29b midpoint (m) This dataset allows reproducibility of the feature importance, SHAP analysis, and magnitude prediction presented in the manuscript.