Predictive Modeling of Basalt-Fiber-Reinforced Bentonite Strength Using Gaussian Process Regression

This study proposes a Gaussian Process Regression (GPR) method for estimating the compressive strength of soil mixtures containing bentonite and basalt fibers. GPR is a preferred probabilistic machine learning approach, especially for limited datasets, due to its high generalization ability and its ability to directly calculate prediction uncertainty. In this study, experimentally obtained bentonite and basalt fiber ratio values were used as input parameters of the model, and unconfined compressive strength (q_u) was determined as the output variable. Model performance was evaluated using the Leave-One-Out Cross-Validation (LOOCV) method. The performance of the proposed GPR model was evaluated with various metrics. Accordingly, the MAE and RMSE values of the model were calculated as 3.667 kPa and 4.756 kPa, respectively, while the R² value was 0.945. The results show that the GPR model provides high prediction accuracy and is a reliable prediction tool for small datasets. Furthermore, the prediction surfaces and uncertainty analyses obtained by the model contributed to a better understanding of the effect of mixture parameters on compressive strength.