Extreme Gradient Boosting Algorithm for Predicting Shear Strengths of Rockfill Materials

For the safe and economical construction of embankment dams, mechanical behaviour rockfill materials used in dam’s shell must be analyzed. The characterization with specified shear strength is difficult expensive due to presence particles greater than 500 mm diameter. This work investigates feasibility using an extreme gradient boosting (XGBoost) computing paradigm estimate materials. To train validate proposed XGBoost model, a total 165 databases obtained from literature are chosen. model was compared against support vector machine (SVM), adaptive (AdaBoost), random forest (RF), K-nearest neighbor (KNN) models described literature. beats SVM, RF, AdaBoost, KNN terms performance evaluation metrics such as coefficient determination (R2), Nash–Sutcliffe (NSE), error root mean square ratio (RMSE) standard deviation measured data (RSR). results demonstrated that has highest prediction (R2 = 0.9707, NSE 0.9701, RSR 0.1729), followed by SVM 0.9655, 0.9639, 0.1899), RF 0.9545, 0.9542, 0.2140), AdaBoost 0.9390, 0.9388, 0.2474) 0.6233, 0.6180, 0.6181). A sensitivity analysis been conducted ascertain impact each investigated input parameter. study demonstrates established for estimating reliable.