Data and code for: Can machine learning predict glacier mass balance for hydrology in unmonitored mountain ranges? A leave-one-region-out test against temperature-index modelling on 184 glaciers using climate reanalysis

Restricted-access data and code record for the manuscript "Can machine learning predict glacier mass balance for hydrology in unmonitored mountain ranges? A leave-one-region-out test against temperature-index modelling on 184 glaciers using climate reanalysis" (Swati Tak; submitted to Journal of Hydrology). Contents: frozen study protocol with SHA-256 freeze record and deviation log; Python 2.7/NumPy code (data retrieval, feature construction, climatology, temperature-index, ridge, random forest, gradient boosting and neural-network models, leave-one-glacier-out and leave-one-region-out cross-validation, statistics, figures, manuscript builder); MERRA-2 daily climate series (NASA POWER) for each WGMS glacier; glacier-year table, cross-validated predictions and result tables; ArcGIS 10.8 map documents and shapefiles (glacier sample, study regions, RGI 6.2 outlines of sample glaciers); figures. WGMS Fluctuations of Glaciers (doi:10.5904/wgms-fog-2026-02-10), RGI 6.2 (doi:10.7265/4m1f-gd79) and Esri World Imagery are not redistributed. Files are available from the author on reasonable request.