Prediction of shear strength parameters for granular soils using machine learning techniques

The shear strength is a significant aspect in many geotechnical problems, obtaining its parameters could be costly and time-consuming especially for granular materials that have large-size particles. In this thesis, an endeavor has been made to examine the use of machine learning techniques to predict the shear strength parameters of granular soils based on some of their index properties. The dataset used to study these techniques is comprised of data obtained from a series of medium-scale direct shear tests in addition to data published in the literature. Several machine learning techniques were examined, and it was evident that many of these techniques could predict the peak friction angle of granular soils with high precision. However, choosing the proper technique is of great importance as well as tuning the model hyperparameters.