Interpretable Symbolic Regression for Reliability-Aware Slope Stability Analysis
Transportation Infrastructure Geotechnology, vol.13, no.5, 2026 (ESCI, Scopus)
- Publication Type: Article / Article
- Volume: 13 Issue: 5
- Publication Date: 2026
- Doi Number: 10.1007/s40515-026-00913-8
- Journal Name: Transportation Infrastructure Geotechnology
- Journal Indexes: Emerging Sources Citation Index (ESCI), Scopus
- Keywords: Factor of safety (FS), Geotechnical modelling, Interpretable machine learning, PySR, Slope stability, Symbolic regression
- Karadeniz Technical University Affiliated: Yes
Abstract
This study developed an interpretable data-driven framework for estimating the factor of safety in slope stability analysis using symbolic regression. A dataset of 10,000 slope cases was assembled, incorporating geotechnical, geometric, hydrological, and reinforcement parameters within physically consistent ranges. Symbolic regression was applied to derive closed-form mathematical expressions that captured nonlinear relationships between input variables and slope stability. The resulting models were evaluated against conventional approaches, including infinite slope and Bishop-type methods, as well as machine learning benchmarks such as linear regression, decision trees, random forests, and neural networks. The symbolic model achieved high predictive accuracy with a coefficient of determination of 0.96 on the test dataset. Sensitivity analysis based on partial derivatives quantified the influence of cohesion, friction angle, slope angle, and pore pressure ratio, confirming physically consistent trends. Probabilistic evaluation using Monte Carlo simulation demonstrated the model’s capability to estimate variability in stability and the probability of failure. The results showed that symbolic regression provided accurate predictions while retaining explicit mathematical structure suitable for engineering interpretation and reliability-based applications.