Advanced mathematical models and their comparison to predict compaction properties of fine-grained soils from various physical properties

This is the authors' abstract. We don't add key points for this paper.

In geotechnical engineering, soil compaction is the practice in which a stress applied to a soil causes densification as air is displaced from the pores between the soil grains. Soil compaction is a vital part of the construction process. This paper describes advanced mathematical models and an innovative solution to predict the compaction properties of fine-grained soils (maximum dry density and optimum moisture content) from various physical properties. Furthermore, it describes and discusses descriptive analytics, inquisitive analytics, and predictive analytics that generate insights and inferences based on the experimental dataset developed in the project. Multiple linear regression (MLR), artificial neural networks (ANNs), and support vector regression (SVR) are employed in the research discussed herein, and the corresponding results are reported and interpreted. A comparative study of the models reveals the one that can identify the best data relationships. Eight neural networks with one or two hidden layers are imposed on the experimental data, and the results show that artificial neural networks in most cases, and support vector regression in some cases, fit the data well with satisfactory results and accuracy.