Supporting dataset for "Machine-learning discovery of hydrolysable ester monomers for degradable thermoset resins"

Supporting dataset for the manuscript "Machine-learning discovery of hydrolysable ester monomers for degradable thermoset resins". Contains: the virtual library of 12,380 candidate diol monomers (SMILES + descriptors); 302 curated xTB hydrolysis-reactivity labels; DFT (wB97X-D3/def2-TZVP, gas phase and IEF-PCM water) calibration results for 6 monomers; ALPB implicit-solvent and oligomer chain-length checks; per-molecule scaffold-split out-of-fold predictions; final model predictions with conformal intervals for the full library; SHAP feature importances; and the Psi4 driver script. Model: XGBoost (300 trees, depth 3, lr 0.05), scaffold OOF MAE = 13.1 kJ/mol, split-conformal q90 = 27.3 kJ/mol.