Machine Learning-based Prediction of Properties of Geopolymer Concrete with Comparative Cost Analysis

GPC is one of the sustainable construction materials. It has a low carbon footprint and good durability. However, predicting mechanical performance and durability related to composition is one of the key challenges compared to Portland cement. This research presents a machine learning–driven predictive framework for evaluating the essential characteristics of geopolymer concrete, including both fresh and hardened properties. In this study, multiple machine learning models are developed, for example, RF, SVG, GB, and ANN. The simulations are assessed using execution metrics, including RMSE, R², and MAE. This combined approach provides engineers, researchers, and field practitioners with a robust tool to support decisionmaking for improving mix design and evaluating the cost-effectiveness of geopolymer concrete. The results determine that the proposed ML models can accurately predict GPC properties with high generalization capability.