Abstract:An accurate evaluation of the clay's properties when mixed with recyclable materials is the end objective of many geotechnical experimental efforts. However, experimental studies need time, cost, and laborious efforts, which could be eliminated using alternative methods. This study concentrated on the prediction of unconfined compressive strength ( UCS ) of marine clay ( MC ) amended with recycled tiles ( RT ). For this aim, four novel hybridized models were developed for estimating UCS of MC modified with RT , with the integration of Adaptive neuro-fuzzy inference system ( ANFIS ), support vector regression ( SVR ), random forest ( RF ), and extreme gradient boosting ( XGB ) with Aquila optimizer ( AO ) algorithm. The results illustrate that all four models contain reliable precision in UCS prediction procedure. Between all models, AO-XGB model has the smallest values of scatter index ( SI ) than others (Excellent performance), with SITrain=0.0247 and SITest=0.02483 . It was clear that the AO-XGB model contains the least uncertainty, with 9.735 and 1.44 for the testing and training dataset, which depicts its greater generalization capability compared to others. All in all, by considering the explanations of results and comparison with published literature, the hybridized AO-XGB model could outperform other models to be practiced in practical usage.

Estimation of unconfined compressive strength of marine clay modified with recycled tiles using hybridized extreme gradient boosting method