Machine learning-based screening and prediction modeling of key influencing factors of slope deformation

Slope deformation, as an important form of geologic hazards, seriously threatens the safety of infrastructure and people's lives and properties. Accurately identifying its key influencing factors and establishing a high-precision prediction model is the key to improve the early warning and management capability of slope disasters. In this paper, based on multi-source monitoring data, correlation analysis, principal component analysis (PCA), recursive feature elimination (RFE) and other methods are used to screen the main influencing factors of slope deformation. On this basis, various machine learning algorithms, including random forest, support vector machine (SVM) and XGBoost, are introduced to construct a slope deformation prediction model, and the model performance is compared and analyzed by error assessment index. The results show that the optimized XGBoost model has the best performance in terms of prediction accuracy and stability. The research in this paper can provide effective technical support for the construction of slope stability assessment and intelligent warning system.