Examining Efficacy of Metamodels in predicting Ground Water Table

摘要

This article examines the capability of Gaussian Process Regression (GPR), Generalized Regression Neural Network (GRNN) and Relevance Vector Machine (RVM) for prediction of Ground Water Table (d_w) at Vellore (India). RVM, GRNN and GPR have been adopted as regression techniques. RVM is a probabilistic model. GRNN approximates any arbitrary function between input and output variables. GPR is a non-parametric model. The developed GPR, RVM and GRNN give the spatial variability of d_w at Vellore. Map of d_w has been also produced by the GPR, RVM and GRNN models. The results show that the developed RVM gives the best model for prediction of d_w at Vellore.