During geotechnical and geophysical site characterisation for large infrastructure projects, significant data volumes are being collected which need to be processed and interpreted. Moreover, the feedback from foundation construction (e.g. pile driving) provides valuable feedback on the actual ground conditions at the site. Due to the limited budgets available for site characterisation and the various sources of uncertainty, the interpretation of ground investigation data and installation records relies on a combination of data from various sources (e.g. in-situ test, laboratory tests and pile load testing), the use of parameter correlations from the literature and expert judgement.

In recent years, modern data science techniques have become increasingly accessible to practicing engineers and researchers and they offer the possibility to improve several aspects of the site characterisation and foundation design process. Machine learning models can be trained on high-quality datasets and expert judgement can also be internalised in the model formulations. In this contribution, the role of data science and machine learning for geotechnical site characterisation and design is discussed based on several example applications using datasets from offshore wind farm projects. The role of data coverage and data quality is discussed as well as the role of geophysical data for interpolating geotechnical point measurements in a quantitative way. Supervised and unsupervised machine learning techniques are explained and illustrated on the provided datasets. The lecture aims to provide engineers an insight into the applicability of machine learning in their daily practice and on the potential issues which may arise when using these methods.

After working in offshore geotechnical consultancy for more than 10 years, Bruno was appointed as Visiting Professor for Offshore Geotechnics at the Geotechnical Laboratory of UGent in 2019. In parallel, he took up a research position at OWI-Lab (VUB/UGent). He is responsible for the Offshore Foundations course and supports the other geotechnical courses in the curriculum. His research interests include geotechnical data management, farm-wide back-analysis of soil structure interaction, the application of data-driven methods to geotechnical design and probabilistic methods.