ABSTRACT
This chapter overviews the current state of data-driven site characterization (DDSC) methods. A 2 × 2 matrix focusing on types of data (numerical or categorical) and uncertainties (geotechnical or geologic) is used to organize different DDSC methods. The strengths and limitations of DDSC methods are highlighted by comparing the methods from a Bayesian perspective. When reviewing the literature on DDSC in geotechnics, it is important to note that many studies and researchers evaluate the proposed methods based on their own validation methods or data, thereby making it challenging to fairly compare the performance of different models. This chapter also discusses benchmark examples or benchmarking for evaluating the strengths and weaknesses of DDSC methods in a balanced and unbiased way.
