Advancing earth science in geotechnical engineering: A data-driven soft computing technique for unconfined compressive strength prediction in soft soil

This study presents a pioneering approach that combines artificial intelligence and a nature-inspired optimization algorithm to predict soil unconfined compressive strength (UCS). The traditional laboratory-based method of UCS measurement, involving soil sample preparation, is time-consuming, labour-intensive, and prone to low accuracy. In this work, we propose a non-destructive soil UCS measurement technique utilizing robust AI-based models based on ensemble learning and hybrid learning techniques. Support vector machine (SVM) coupled with particle swarm optimization (PSO), extreme gradient boost (XGB), K-nearest neighbour (KNN), and nature-inspired optimization algorithm-based six hybrid ANFIS models, employing input features from experimental data, were adopted for UCS prediction. Model performance was assessed using standard metrics such as root mean square error, mean absolute error, variance account factor (VAF), expanded uncertainty (U₉₅), and coefficient of determination (R²) between predicted and actual unconfined compressive strength. The study employed 274 data points generated in our laboratory. Sensitivity analysis and Pearson correlation techniques were employed to select relevant elements as input features. Fine content, coarse content, liquid limit, plastic limit, plasticity index, and cohesion of soil were identified as the most effective configurations for accurate soil UCS predictions. XGB demonstrated the highest prediction efficiency in the training and testing phase, achieving an impressive R² of 99.2 and 96.8%, respectively. The results also emphasize the importance of the selected features. The experimental validation accuracy of 97% for the developed XGB model, whose data were not used during model calibration and verification, confirmed the generalization capability of the models. This study provides valuable insights for policymakers and industry stakeholders, facilitating optimized soil unconfined strength management practices.