Design of Mixtures and Manufacture of Self-Compacting Concretes with Recycled Aggregates (Eco-Concretes): Prediction of Compressive Strength Using Machine Learning Models

The aim is to predict the compressive strength of self-compacting concrete with recycled aggregates by using machine learning methods. Various ensemble methods and generalized additive models were applied to a sample of 515 research articles divided into training, validation and testing. Compressive strength was considered as output variable and SCC components as input variables. In addition, several metrics were used to evaluate the models in terms of their prediction. The results indicate that the Random Forest and Gradient Boosting models are capable of accurately predicting compressive strength, and that cement and water are the variables with the greatest impact on the prediction.