Developing a RFR Model to forecast the Rock-Socketed Piles' Ultimate Bearing Capacity

This investigation proposes a novel scheme to address the multifaceted challenges in anticipating the Ultimate Bearing Capacity of rock-socketed piles, which is crucial in rocky terrain foundation engineering. By integrating the Random Foreset Regression (RFR) model with two innovative meta-heuristic optimization tactics, namely the Reptile Search and the Wild Geese Schemes, this research offers unique contributions to forecasting modeling and data analysis. These schemes enhance the accuracy and efficacy of ML patterns through soft computing, aiming to fill a gap in the existing body of knowledge. The novelties of the present study are displayed by the original integration of meta-heuristic optimization tactics with ML tactics to improve the accuracy and efficacy of the prediction of Ultimate Bearing Capacity of rock-socketed piles, moving the state-of-the-art in foundation engineering. Based on the thorough assessment using performance indicators like R^2, RMSE, MSE, NRMSE, and the U95 index, the RFWG model, as proposed, shows outstanding performance, reaching a maximum R^2 of 99.8% and a minimal RMSE of 513.86. In contrast, the single RFR model registered the weakest performance, with 97.8% for R^2 and 1625.02 for RMSE. These outcomes emphasize the effectiveness of RFWG in real-world scenarios and its promise in driving forward predictive modeling and data analysis within the foundation engineering context.