Predictive Modeling for Compressive Strength in Sustainable Concrete Using Machine Learning Techniques

The large-scale fly ash and slag integration in concrete has not only aided in reducing the environmental impact of cement, but also contributed to economic development by reducing waste, cutting carbon emissions, conserving resources, and offering cost-effective, sustainable construction solutions. This aligns with the global shift towards green building practices. However, the optimal mix design of such green concrete continues to remain a significant challenge since it requires optimizing component proportions to achieve several desired attributes of concrete. This study focuses on optimizing concrete mix design by forecasting a crucial variable: 28-day compressive strength. It is analyzed using seven input factors: cement, slag, fly ash, water, superplasticizer (sp), coarse aggregate, and fine aggregate. A dataset comprising 103 data points from the literature is employed to investigate the chosen input and output variables. Various machine learning methods, such as linear regression, logistic regression, decision trees, random forests, support vector machines, k-nearest neighbors, k-means clustering, neural networks, and Gaussian process regression, are evaluated on the gathered dataset. The performance metrics from all machine learning models are compared to evaluate their efficiency. The results from the analysis indicate that Gaussian Process Regression (GPR) demonstrates exceptional accuracy in forecasting 28-day compressive strength, making it a robust choice for precision–critical applications in concrete mix design.