A comparative study of machine learning algorithms for the prediction of compressive strength of rice husk ash-based concrete

The cementitious behavior of Rice Husk Ash (RHA) has caused its possible addition as a replacement material for cement which has been proven to influence the strength of concrete. In this study, Machine Learning (ML) algorithms have been used to predict the compressive strength of RHA-based concrete in a shorter period without any errors. In this regard, six different ML techniques, i.e., Linear Regression, Decision Tree, Gradient Boost, Artificial Neural Network, Random Forest and Support Vector Machines, have been employed to predict the compressive strength using twelve input features and 462 data points. The performances of models have been checked using errors, Pearson correlation coefficient ( R 2 ), Taylor’s diagram, box plots and Sensitivity analysis. The outcome of this study indicated that the Decision Tree, Gradient Boost, and Random Forest models had provided better results ( R 2 > 0.92) than the other algorithms in terms of minimal errors and high accuracy in predicting compressive strength. The sensitivity analysis indicated that the specific gravity of RHA and water–cement ratio significantly (more than 95%) impact the compressive strength of the RHA-based concrete in contrast to the other parameters.