Machine learning approach for the estimation of the maximum dry density and optimum moisture content of the stabilised soils

This research presents a machine learning framework to assess the maximum dry density and optimal moisture content of cement-stabilised soils. Two predictive models random forest regression and Gaussian process regression were created and refined by red-tailed hawk optimisation and sand cat swarm optimisation, yielding hybrid models. The random forest regression model, optimised by red-tailed hawk optimisation, exhibited exceptional performance, attaining coefficients of determination of 0.988 in training, 0.979 in validation, and 0.978 in testing for predicting maximum dry density, with corresponding root mean square error values of 30.72, 24.64, and 42.11, respectively. The model attained coefficients of determination of 0.993, 0.992, and 0.978. The random forest regression model augmented by red-tailed hawk optimisation demonstrated superior accuracy and reliability, serving as an effective instrument to diminish dependence on laborious laboratory testing and raise the efficiency of construction methodologies.