Comparison of machine learning techniques to predict the compressive strength of concrete and considerations on model generalization

Abstract The compressive strength of concrete is an essential property to ensure the safety of a concrete structure. However, estimating this value is usually a laborious and uncertain process since the mix design is based on empirical methods and its confirmation in the laboratory demands time and resources. In this context, this work aims to evaluate Machine Learning (ML) models to predict the compressive strength of concrete from its constituents. For this purpose, a dataset from the literature was used as input to four ML models: Extreme Gradient Boosting (XGBoost), Support Vector Regression (SVR), Artificial Neural Networks (ANN) and Gaussian Process Regression (GPR). The accuracy of the models was evaluated through 10-fold cross-validation, and quantified by R2, Mean Absolute Error (MAE), and Root-Mean-Square Error (RMSE) metrics. Subsequently, a new dataset was put together with mixtures from the literature and used to validate the previous models. In the model creation step, all algorithms obtained similar and positive results, with MAE between 1.96-2.26 MPa and R2 varying from 0.79 to 0.83. However, in the validation step, the accuracy of the models dropped sharply, with MAE growing to 3.04-4.04 MPa and R2 decreasing to 0.37-0.59. ANN and GPR showed the best results, while SVR had the worst predictions. This work showed that ML tools are promising techniques to predict the compressive strength of concrete. However, care must be taken with the input data to guarantee that models are not overfitted to a given region, set of materials, or type of concrete. Resumo A resistência à compressão do concreto é uma propriedade essencial para garantir a segurança de uma estrutura. No entanto, estimar este valor é atualmente um processo trabalhoso e impreciso, uma vez que o a dosagem é baseada em métodos empíricos e sua confirmação em laboratório demanda tempo e recursos. Nesse contexto, este trabalho tem como objetivo avaliar modelos de Aprendizado de Máquina (ML) para predizer a resistência à compressão do concreto a partir de seus componentes. Para tanto, um banco de dados da literatura foi utilizado como entrada para quatro modelos de ML: Extreme Gradient Boosting (XGBoost), Regressão de Vetor de Suporte (SVR), Redes Neurais Artificiais (ANN) e Processo Gaussiano de Regressão (GPR). A precisão dos modelos foi avaliada por meio de validação cruzada (10-fold) e medida com as métricas de R2, Erro Médio Absoluto (MAE) e a Raiz do Erro Quadrático Médio (RMSE).