Evolutionary polynomial regression (EPR) applications in geotechnical engineering: A comprehensive survey, uncertainty analysis, and physical interpretation of models

This article presents a systematic review of 98 peer-reviewed studies on the application of Evolutionary Polynomial Regression (EPR) in geotechnical engineering, integrating a descriptive approach and a critical evaluation of the publications, with the aim of highlighting the evolution of these algorithms and identifying opportunities for future advancements. Additionally, original data and independent analyses are included both to demonstrate what is being done at the forefront of the field and to support the arguments and directions presented. Despite the growing interest of the geotechnical community in applying EPR to geotechnical engineering problems, no systematic review has yet been conducted to undertake a critical evaluation of the basic algorithm configurations, examine the primary outputs and predictive parameters, or assess model accuracy. This study reveals the broad versatility of EPR within geotechnical engineering, with most applications focused on foundations, soil mechanics, and soil stabilization. Comprehensive summary tables are provided, detailing the main features of the models across twelve geotechnical themes. The review also highlights critical gaps, particularly the frequent lack of strategies to mitigate overfitting and the limited use of cross-validation methods. This systematic review underscores the potential of EPR for developing predictive models in geotechnical engineering and offers future directions for refining and enhancing this technique in the field. Systematic review of 98 articles on EPR applications in geotechnical engineering, integrating description and critical evaluation of the publications. Versatility of EPR in geotechnics, with a focus on applications in foundations, soil mechanics, and soil stabilization; Analysis of the main model characteristics and identification of key inputs and outputs by subarea of geotechnical engineering; Identified methodological gaps: limited use of multi‑objective strategies, cross‑validation, and sensitivity analysis.