Multivariate Adaptive Regression Splines Models for Analyses of Diaphragm Wall and Drilled Shafts — Numerical Case Studies

Abstract

Analyses of diaphragm wall deflections and undrained side resistance for drilled shafts rely on the use of empirical methods due to an inadequate understanding of the physical phenomena involved in these highly nonlinear, multivariate problems. Various advanced computational learning tools, such as artificial neural networks (ANN), have been increasingly used. However, neural networks have been criticized for the long training process. This paper explores the use of a fairly simple, nonparametric regression algorithm known as multivariate adaptive regression splines (MARS) to mathematically model multivariate, nonlinear problems through two numerical case studies. The main advantages of MARS are highlighted. First, the MARS methodology is described. MARS models of diaphragm wall deflections and undrained side resistance for drilled shafts are then presented.