The scour phenomenon is highly complex, and its precise prediction remains a considerable challenge for hydraulic researchers. Traditionally, most of the studies relay on empirical approaches for scour prediction. However, with the rapid development of infrastructure and the increasing number of bridges, these empirical models are inadequate in providing precise scour prediction across diverse field conditions. Consequently, there is a rising requirement to adopt advanced Machine Learning (ML) based methodologies to achieve more precise and computationally efficient scour prediction. The primary objective of this study is to propose an alternate model to traditional scour prediction methods by employing ML-based tools, specifically Artificial Neural Networks (ANN), Support Vector Machines (SVM), and the Ensemble Tree Method. These ML-based models are capable of handling complex and nonlinear problems. In addition to simulation work, a detailed experimental investigation has been carried out for a single cylindrical pier. The results of the present study, along with existing literature data, have been utilized for training and testing of ANN, SVM, and Ensemble Tree models. To evaluate the performance of the proposed models, an in-depth statistical analysis has been carried out. The findings of this study highlight that all three models significantly outperform traditional methods, demonstrating their effectiveness in improving scour depth predictions

-