Spatial variability of rock depth using adaptive neuro-fuzzy inference system (ANFIS) and multivariate adaptive regression spline (MARS)

Abstract

This article adopts adaptive neuro-fuzzy inference system (ANFIS) and multi adaptive regression spline (MARS) for prediction of spatial variability of reduced level of rock depth (d) in Bangalore. The database consists of 652 d values spread over a 220 km2 area of Bangalore. In two-dimensional analysis, the function \(d = f (x_{\text{1}} ,x_{\text{2}} )\), where x1, and x2 are the latitude and longitude of a point corresponding to d value, is to be approximated with which d value at any half space point in Bangalore can be determined. The input variables of ANFIS and MARS are x1 and x2. Map of d has been also produced by developed ANFIS and MARS models. The comparison between the MARS and ANFIS models demonstrates that the MARS model is superior to the ANFIS model.