<p>This study presents a systematic data-driven approach for predicting the compressive strength (CS) of fly ash-based geopolymer mortars using three statistical modeling techniques Linear Regression (LR), Multiple Linear Regression (MLR), and Nonlinear Regression (NLR). The primary objective is to address the inherent complexity in geopolymer mortar mix designs, where multiple interdependent variables such as fly ash content, SiO₂ and Al₂O₃ percentages, sand content, liquid-to-binder ratio (l/b), curing time, and specimen age influence strength development. A robust dataset of 280 experimentally validated samples was employed, partitioned into training (70%), testing (15%), and validation (15%) subsets. Each model was calibrated using least squares optimization, and evaluated through standard performance metrics such as R², RMSE, and MAE. Among the models, the NLR model achieved the highest predictive performance (R² = 0.9483, RMSE = 5.14&#xa0;MPa for training; R² = 0.937 for both testing and validation), effectively capturing the nonlinear interdependencies among input variables. MLR and LR demonstrated acceptable but lower predictive accuracies and greater residual dispersion. Residual error plots further substantiated the NLR model’s robustness, with minimal deviation across all datasets. This work contributes novel insights by developing a nonlinear regression framework tailored specifically for geopolymer mortar distinct from more commonly studied concrete systems thereby enhancing the predictive design process for sustainable construction materials. Practically, the developed models offer a valuable framework for performance-based mix optimization of geopolymer mortars, significantly reducing the need for extensive laboratory experimentation. By accurately predicting compressive strength based on mix design and curing parameters, these models facilitate faster and cost-effective decision-making during the material development phase.</p>