Data, code, and supplementary archive for a cross-database natural-carbonation ML audit for low-clinker binders

Computational reproducibility archive supporting a Cement and Concrete Research manuscript on chemistry-guided machine learning, grouped residual-margin calibration, external transfer stress testing, and virtual low-clinker screening for natural-carbonation durability. The public release includes analysis scripts, generated aggregate tables, manuscript figures, a supplementary-material file, reproducibility notes, and source-data guidance. Raw third-party datasets are not redistributed.