Slope Stability Prediction Using k-NN-Based Optimum-Path Forest Approach

Slope instability can lead to catastrophic consequences. However, predicting slope stability effectively is still challenging because of the complex mechanisms and multiple influencing factors. In recent years, machine learning (ML) has received great attention in prediction due its strong nonlinear ability. this study, an optimum-path forest algorithm based on k-nearest neighbor (OPFk-NN) was used predict slopes. First, 404 historical slopes with failure risk were collected. Subsequently, dataset train test randomly divided training sets, respectively. The hyperparameter values tuned by combining ten-fold cross-validation grid search methods. Finally, performance proposed approach evaluated accuracy, F1-score, area under curve (AUC), computational burden. addition, results compared other six ML algorithms. showed that OPFk-NN had a better performance, AUC, burden 0.901, 0.902, 0.957 s, Moreover, failed cases be accurately identified, which highly critical prediction. angle most important influence results. Furthermore, engineering application overall predictive model consistent factor safety value This study provide valuable guidance for analysis management.