Accepted: 12 February 2024 / Published online: 22 February 2024 | Ahsan Rabbani, Jan Afzal Muslih, Mukul Saxena, Santosh Kalyanrao Patil, Bharat Nandkumar Mulay, Mohit Tiwari, A Usha, Sunita Kumari, Pijush Samui
The paper "Utilization of Tree-Based Ensemble Models for Predicting the Shear Strength of Soil" by Ahsan Rabbani et al. focuses on developing and validating tree-based ensemble models to predict the shear strength (SS) of soil. The study uses a dataset of 249 soil samples and six controlling variables: depth, clay proportion, loam proportion, sand proportion, plastic limit, and plastic index. Five tree-based models—extra tree regressor (ETR), decision tree regressor (DTR), ridge regression (RR), linear regression (LR), and Bayesian ridge regressor (BRR)—were developed and evaluated using six performance indices. The results indicate that ETR outperforms the other models, as confirmed by rank analysis and sensitivity analysis. The feature importance plot revealed that sample depth is the strongest contributing factor to SSS prediction in the best model. The study also utilized various diagnostic plots to assess the model's performance, including learning curves, outlier distance detection plots, feature importance plots, prediction error plots, and residual plots. The findings highlight the effectiveness of ETR in predicting soil shear strength without the need for costly laboratory testing.The paper "Utilization of Tree-Based Ensemble Models for Predicting the Shear Strength of Soil" by Ahsan Rabbani et al. focuses on developing and validating tree-based ensemble models to predict the shear strength (SS) of soil. The study uses a dataset of 249 soil samples and six controlling variables: depth, clay proportion, loam proportion, sand proportion, plastic limit, and plastic index. Five tree-based models—extra tree regressor (ETR), decision tree regressor (DTR), ridge regression (RR), linear regression (LR), and Bayesian ridge regressor (BRR)—were developed and evaluated using six performance indices. The results indicate that ETR outperforms the other models, as confirmed by rank analysis and sensitivity analysis. The feature importance plot revealed that sample depth is the strongest contributing factor to SSS prediction in the best model. The study also utilized various diagnostic plots to assess the model's performance, including learning curves, outlier distance detection plots, feature importance plots, prediction error plots, and residual plots. The findings highlight the effectiveness of ETR in predicting soil shear strength without the need for costly laboratory testing.