<p>Geopolymer concrete is an environmentally friendly substitute for ordinary portland cement concrete, due to its superior strength and decreased environmental emission. This study employs eight machine learning algorithms (e.g., LR, DT, RF, AdaBoost, SVR, XGBoost, LightGBM, and CatBoost); a dataset sourced from 55 experimental studies comprising 1632 data points was utilized to predict the compressive strength of fly ash-based geopolymer concrete, leading to improved geopolymer concrete structure design and performance. Research methodology involves several stages (e.g., database setup, data collection, exploratory data analysis, removal of outliers, data standardization, and model development). Exploratory data analysis provides statistical information for the input and output parameters (e.g., units, mean, median, minimum, maximum, variance, standard deviation, kurtosis, and skewness). The model's performance and accuracy were evaluated using six metrics (e.g., R², Adj. R², MAE, RMSE, MSE, and SMAPE). During the testing phase, XGBoost demonstrated superior performance, achieving an R² value of 0.95, MAE of 1.84, MSE of 11.76, and RMSE of 3.43. Based on six evaluation metrics, XGBoost outperformed all other models, followed by RF &gt; CatBoost &gt; DT &gt; LightGBM &gt; SVM &gt; AdaBoost &gt; LR. Through error analysis techniques, including fluctuations, distribution, and maximum error evaluations, offer insights into the accuracy and reliability of these models. Shapley additive explanations (SHAP) enhance interpretability by quantifying feature importance through ranked values and absolute mean impacts, revealing curing temperature as the most influential factor in this research. Furthermore, the application of machine learning techniques in the construction industry promotes efficiency and sustainability, fostering an environmentally conscious and resilient built environment.</p>