TY - JOUR
TI - Prediction of rock slope failure based on multiple machine learning algorithms
AU - Mnzool Mohammed
JN - Thermal Science
PY - 2024
VL - 28
IS - 6
SP - 4907
EP - 4916
PT - Article
AB - Slope failures have the potential to seriously jeopardize access to  sustainable development since they cause numerous casualties as well as  disastrous effects on society and the economy. It is imperative to use  precise operable computational designs in this case. This study examined  the efficacy of five distinct machine learning models, namely support vector  machines, decision trees, gradient boost machine learning, and random  forest, in predicting the slope safety factors. This article's primary goal  is to assess and improve the different machine learning-based analytical  representations in relation factor of safety computations. The genetic  algorithm mimics the processes of growth, hybridization, and mutagenesis  found in the expected collection and inherent procedures to resolve the  hyperparameters of machine learning algorithms. A total of 217 cases were  collected in order to train and evaluate these models. Multiple convergence  analysis is also used to study the independence of individual  characteristics. The assessed methods' competence was assessed through the  application of diverse performance assessment indicators. The various  classifiers function satisfactorily for slope failure inquiry, according to  the evaluation and comparison of the data. Random forest was found to be the  best classification method for slope failure prediction, with an accuracy of  91%.