Optimizing soil settlement/consolidation prediction in finland clays: machine learning regressions with bayesian hyperparameter selection

This study focuses on optimizing soil settlement and consolidation prediction in Finland clays using machine-learning regressions with Bayesian hyperparameter selection. Specifically, the study aims to predict the pre-consolidation stress (sp) using an Extra Trees Regressor (ETR) model. Root mean square error (RMSE) was used as a performance metric to evaluate the model's accuracy c. Several machine-learning models were trained and tested, and the ETR was found to have the highest testing R2 value of 0.7614. Bayesian hyperparameter selection was then used to optimize the model's performance, and the area under the curve (AUC) score was used to evaluate the optimization process. The optimization process yielded a convergence plot that started with a hyperparameter value, and the AUC score reached a maximum of 0.95. Overall, this study provides valuable insights into the application of machine learning techniques and Bayesian optimization in predicting pre-consolidation stress in Finland clays.