The determination of rock depth is an important task in geotechnical and geological engineering. This article examines the capability of simple kriging, ordinary kriging, Relevance Vector Machine (RVM) and Minimax Probability Machine Regression (MPMR) for prediction of rock depth at any point in Vellore(India). For simple and ordinary kriging, semivariogram model has been developed. RVM is developed based on the Bayesian theory. MPMR is a probabilistic model. Inputs of the models are latitude (Lx) and longitude (Ly). A comparative study has been carried out between the developed simple kriging, ordinary kriging, RVM and MPMR models. The developed simple kriging, ordinary kriging, RVM and MPMR give rock depth maps of Vellore. The developed RVM and MPMR give better performance than the simple and ordinary kriging.