Machine Learning for Predicting Tunnel-Induced Settlements: From PHD Research to an Interactive Educational Platform

This article explores the application of Machine Learning (ML) techniques in geotechnical engineering, specifically focusing on the prediction of settlements in tunnel projects. As the use of ML in geotechnics continues to evolve, the integration of robust data analytics tools within collaborative platforms offers new opportunities for precision and efficiency. We present a case study in which data from tunnel construction projects are processed, cleaned, and analyzed to predict ground settlements, a critical factor in tunnel design and safety. The study outlines the preparation of geotechnical data, including extraction, cleaning, and feature engineering, followed by the application of ML algorithms for predictive modeling. Key steps in the methodology, such as variable selection, data visualization, and model evaluation, are discussed, with an emphasis on the tools available within the platform for seamless integration of geotechnical data and machine learning workflows. Furthermore, the article highlights the iterative nature of model training, optimization, and real-time prediction, showcasing the platform's ability to provide a user-friendly interface for engineers to interact with advanced ML models. The results demonstrate how ML can enhance the accuracy of settlement predictions and contribute to more reliable decision-making in tunnel design, ultimately advancing the role of data-driven methodologies in the field of geotechnics.

Keywords: Database, Data visualisation, Statistics.