Comparing the performance of machine learning models for predicting the compressive strength of concrete

Abstract

Abstract This study aimed to investigate and compare the performance of different machine learning models in predicting the compressive strength of concrete using a data set of 1234 compressive strength values. The predictive variables were selected based on their relevance using the SelectKBest method, resulting in an analysis of eight and six predictive variables. The evaluation was conducted through linear correlation studies via simple linear regression and non-linear correlation studies using support vector regression (SVR), gradient boosting (GB), and artificial neural networks (ANN). The results showed a coefficient of determination (R²) = 0.85 and a root mean square error (RMSE) = 30.9051 MPa for SVR, R² = 0.90 and RMSE = 25.5979 MPa for GB, and R² = 0.87 and RMSE = 5.781 MPa for ANN. The comparison between the machine learning methods revealed significant differences. For instance, GB stood out with a higher R² value, demonstrating its remarkable ability to explain the variability in the data. Conversely, ANN showed the lowest RMSE value, indicating notable accuracy in the predictions. The choice between these approaches depends on considerations regarding the balance between explainability and accuracy. While GB provides a more in-depth understanding of the relationship between variables, ANN stands out for the accuracy of its predictions.