4. Conclusion

This article takes an in-depth look at ML applications in geotechnical engineering, illustrating its application to the use case of predicting settlement induced by tunnel excavation. It stresses the importance of prioritizing data quality over quantity, an imperative in a field where data from construction sites is often heterogeneous, noisy or incomplete. Although time-consuming and demanding, the data preparation phase is an essential investment in obtaining reliable, usable results.

Analyses show that ML models, particularly set methods such as XGBoost or RF, enable accurate predictions and robust generalization, while adapting to the demands of complex projects thanks to rapid training and the ability to easily integrate new data. These approaches offer a decisive advantage in optimizing current practices and anticipating specific geotechnical challenges....