Public concrete datasets often contain duplicate records, coupled variables, and cross-source distribution shifts, which may lead to overly optimistic model evaluation. Based on a deduplicated UCI high-performance concrete dataset (1005 samples), this study develops a leakage-controlled data-driven workflow with applicability-domain assessment. TabPFN, SHAP, and NSGA-II are used for compressive strength prediction, model-response attribution, and scenario-based candidate mix screening, respectively. Model evaluation follows a unified data split, inner training-set cross-validation, and an independent test-set protocol. In addition, 502 non-overlapping records from the Mendeley PCC dataset are used as an external benchmark to examine cross-source transferability and sensitivity to distribution shift. The results show that TabPFN achieves the highest R² and the lowest RMSE, MAE, and MAPE on the internal UCI test set, with values of 0.953, 3.744 MPa, 2.265 MPa, and 7.580%, respectively; however, its advantage over strong baselines such as CatBoost is limited. On the external Mendeley PCC dataset, TabPFN remains competitive, with R², RMSE, and MAE values of 0.490, 15.175 MPa, and 11.457 MPa, respectively, but its performance is close to that of random forest, XGBoost, and CatBoost. The 5NN applicability-domain stratification shows that external samples located within the 95% 5NN applicability domain achieve improved performance (R² = 0.634 and RMSE = 12.367 MPa), suggesting that external prediction errors are associated with the distance from the source-domain distribution. SHAP results indicate that cement, ground granulated blast-furnace slag, curing age, and water are the main attribution variables in the model output; their response directions should be interpreted as statistical attributions rather than material causal mechanisms. The Pareto candidate mixes generated by NSGA-II satisfy basic engineering constraints. Nevertheless, because the external benchmark reveals sensitivity to cross-source distribution shift, the resulting mix proportions should be treated as pre-experimental screening candidates rather than engineering-validated low-GWP concrete mix proportions.

Public geopolymer concrete (GPC) databases can support retrospective screening of records with lower reported CO₂ emissions under strength constraints, but reliable ranking also requires correct response matching and stable threshold-based eligibility. Using 274 records, this study reconstructed the strength–CO₂ correspondence from 19 mix proportion and preparation fields, yielding 273 mix design identifiers (MixIDs). Eight regression algorithms were evaluated under MixID-grouped validation, and CatBoost with derived descriptors was retained as the primary model. It achieved R² = 0.969 and RMSE = 2.096 MPa on the fixed test set; across 30 repeated holdouts, mean RMSE was 2.390 ± 0.234 MPa. Because observed strength was available for every record, the model served as an internal consistency filter rather than a predictor of untested formulations. At 40, 50, and 60 MPa, 176, 65, and four records met the observed strength and mean prediction criteria, while 174, 62, and three records also satisfied the empirical fifth-percentile (Q05) criterion. Q05 was treated as a descriptive measure of partition sensitivity, and the 60 MPa result was regarded as exploratory. A CO₂ assignment negative control showed weak agreement between row-wise and key-matched values (Pearson r = 0.065) and low Pareto-set overlap (Jaccard = 0.045–0.105). Thus, prediction accuracy alone does not ensure valid lower-CO₂ ranking. The workflow provides an auditable basis for screening observed records within the database domain.