A Machine Learning Perspective on Slope Stability: Evaluating Logistic Regression and Random Forest

Slope stability analysis serves as an essential method to stop landslides and secure infrastructure together with human settlements. The environmental methods of analysis depend upon outdated empirical models along with deterministic methods that show weakness when facing multifaceted geotechnical settings. A machine learning analysis of slope stability prediction is described in this research which evaluates the effectiveness between Logistic Regression models and Random Forest models. Training occurred using a dataset that consisted of geological elements together with geotechnical and environmental factors. Random Forest delivers better accuracy and stronger robustness than Logistic Regression despite Logistic Regression being capable of producing interpretive decision borders. The model evaluation consisted of assessing accuracy as well as precision and recall and F1-score metrics. The analysis proves the effectiveness of applying machine learning approaches to improve slope stability prediction outcomes. This research demonstrates the need for data-based strategies in geotechnical engineering and recommends more advanced machine learning techniques to enhance future risk assessment capabilities. Multiple studies about this topic have implemented different research methods. Slope characteristics and stability conditions can be investigated with machine learning (ML) algorithms by processing monitoring and testing data. The authors have attempted to provide a thorough review of the literature involving ML techniques in slope stability analysis within this paper. The research demonstrated that scientists mostly depended on data-driven methods which used minimal input factors while demonstrating the ML techniques' ability to effectively forecast slope failure outcomes. SVM and RF represented the predominate ML models researchers relied upon.