In the geotechnical engineering field, it is required to anticipate the Factor of Safety (FOS) in slope stability precisely in order to assess the possibility of slope failure and guarantee infrastructure safety. This research utilizes a thorough slope stability dataset to inspect how well six tree-based regression models—Decision Tree, Random Forest, Extra Trees, AdaBoost, Gradient Boosting, and XGBoost—predict the FOS. With the target of predicting the continuous FOS value, the dataset covers 10,000 samples with eight vital geotechnical parameters and one categorical reinforcement feature. Using performance metrics like RMSE, MAE, R2 score, and execution time, a modified study was executed. The most significant factors affecting slope stability were also resolved using feature importance analysis. The Extra Trees Regressor performs finer than other models in terms of predictive accuracy, according to the results, while cohesion, internal friction angle, slope angle, and pore water pressure ratio decrease.
Introduction
Slope stability analysis is essential in geotechnical engineering for preventing landslides and ensuring structural safety. The Factor of Safety (FoS) quantifies slope stability, with values ≥1 indicating stable slopes. Traditional analytical methods often rely on deterministic assumptions, limiting their ability to capture complex nonlinear relationships among geotechnical factors. Machine learning, particularly tree-based regression models, offers robust, interpretable, and accurate alternatives for predicting FoS.
This study evaluates six tree-based regressors—Decision Tree, Random Forest, Extra Trees, AdaBoost, Gradient Boosting, and XGBoost—using a large synthetic dataset with 10,000 samples and key geotechnical parameters (e.g., cohesion, friction angle, slope height). Models are assessed based on accuracy (RMSE, MAE, R²), computational time, and feature importance, bridging traditional geotechnical methods with data-driven approaches.
A detailed literature review highlights prior research employing machine learning for slope stability, noting limitations like small datasets, limited features, and classification focus. In contrast, this work focuses exclusively on regression using a large synthetic dataset with both continuous and categorical variables.
The methodology includes data preprocessing (encoding, scaling, train-test split), model training and evaluation, and interpretation of influential geotechnical features. The study underscores the advantages of ensemble tree methods in handling nonlinearity, improving prediction accuracy, and offering insights into key factors affecting slope stability.
Conclusion
This study illustrates the effectiveness of tree-based regressors in precisely predicting the Factor of Safety for slope stability analysis. Among all the models, the Extra Trees Regressor consistently reached the highest implementation over RMSE, MAE, and R² metrics. Feature significance study announced that cohesion, internal friction angle, slope angle, and pore water pressure ratio were the most influential geotechnical factors. The outcomes emphasize the perspective of integrating machine learning with geotechnical engineering for apocalyptic modelling. Future work can expand on this by incorporating additional soil conditions and real-time monitoring data for upgraded conception.
References
[1] W. Zhang, H. Li, L. Han, L. Chen, and L. Wang, \"Slope stability prediction using ensemble learning techniques: A case study in Yunyang County, Chongqing, China,\" Journal of Rock Mechanics and Geotechnical Engineering, vol. 14, no. 4, pp. 1089–1099, 2022.
[2] A. Kainthola, V. H. R. Pandey, G. Kushwaha, and V. Yadav, \"Role of Feature Importance in Geomechanical Classification of Rock Slopes,\" Deep Resources Engineering, pp. 100165, 2025.
[3] Z. Xu, X. Wang, L. Guo, and T. Yu, \"Stability Evaluation of Slope Based on Global Sensitivity Analysis,\" Complexity, vol. 2024, no. 1, pp. 2333859, 2024.
[4] P. Ragam, N. Kushal Kumar, J. E. Ajith, G. Karthik, V. K. Himanshu, D. Sree Machupalli, and B. Ramesh Murlidhar, \"Estimation of slope stability using ensemble-based hybrid machine learning approaches,\" Frontiers in Materials, vol. 11, pp. 1330609, 2024.
[5] F. Ahmad, X. Tang, J. Hu, M. Ahmad, and B. Gordan, \"Improved prediction of slope stability under static and dynamic conditions using tree-based models,\" parameters, vol. 1, no. 3, 2023.
[6] A. K. Sahoo, D. P. Tripathy, and S. Jayanthu, \"Application of machine learning techniques in slope stability analysis: A comprehensive overview,\" Journal of Mining and Environment, vol. 15, no. 3, pp. 907–921, 2024.
[7] M. H. Kadkhodaei, E. Ghasemi, and M. H. Fazel, \"Slope stability prediction of circular mode failure by machine learning models based on Bayesian Optimizer,\" Journal of Mountain Science, vol. 22, no. 4, pp. 1482–1498, 2025.
[8] B. Zerouali, N. Bailek, A. Tariq, A. Kuriqi, M. Guermoui, A. H. Alharbi, D. S. Khafaga, and E.-S. M. El-Kenawy, \"Enhancing deep learning-based slope stability classification using a novel metaheuristic optimization algorithm for feature selection,\" Scientific Reports, vol. 14, no. 1, pp. 21812, 2024.
[9] D. K. Yadav, S. Chattopadhyay, D. P. Tripathy, P. Mishra, and P. Singh, \"Enhanced slope stability prediction using ensemble machine learning techniques,\" Scientific Reports, vol. 15, no. 1, pp. 7302, 2025.
[10] K. C. Onyelowe, A. M. Ebid, S. Hanandeh, and V. Kamchoom, \"Evaluating the slope behavior for geophysical flow prediction with advanced machine learning combinations,\" Scientific Reports, vol. 15, no. 1, pp. 6531, 2025.