Related Experiment Video
Updated: Jun 14, 2026
08:27
Image Recognition and Parameter Analysis of Concrete Vibration State Based on Support Vector Machine
Published on: January 5, 2024
Prediction of soil shear strength using hybrid machine learning approaches for performance and interpretability
Muhammad Suliman1, Maaz Khan2,3, Touqeer Ali Rind4,5
1Department of Civil and Environmental Engineering, Nagaoka University of Technology, Nagaoka, Niigata, Japan.
Scientific Reports
|June 12, 2026
Summary
Machine learning models accurately predict soil shear strength, a key factor in construction stability. Random Forest and Support Vector Machine models showed the best performance, offering efficient alternatives to traditional testing.
Area of Science:
- Geotechnical Engineering
- Machine Learning Applications
Background:
- Soil shear strength is critical for construction stability.
- Conventional lab testing for soil shear strength is costly and time-consuming.
Purpose of the Study:
- To predict soil shear strength using machine learning (ML) models.
- To evaluate the performance of ML models including Multiple Linear Regression (MLR), Support Vector Machine (SVM), Random Forest (RF), and Multi-Expression Programming (MEP).
Main Methods:
- Utilized geotechnical parameters from 202 soil samples.
- Developed and analyzed four predictive ML models.
- Evaluated model performance using R², RMSE, MAE, residual analysis, and Taylor diagrams.
Main Results:
- Random Forest (RF) exhibited the best predictive performance (training R²=0.9527, testing R²=0.8578).
- Support Vector Machine (SVM) also showed impressive predictive capabilities.
- SHAP analysis identified liquid index, moisture content, and plasticity index as key influencing factors.
Conclusions:
- Modern ML models like RF and SVM are efficient for modeling the nonlinear behavior of soil characteristics.
- ML models offer a viable and efficient alternative to conventional methods for predicting soil shear strength.