Rational Kriging

Kriging is a technique for multivariate interpolation of arbitrarily scattered data. It is originated from some mining-related applications, which is developed into the field of geostatistics by the pioneering work of Matheron (1963). It has now become a prominent technique for function approximation and uncertainty quantification in spatial statistics (Cressie, 2015), computer experiments (Santner et al., 2003), and machine learning (Rasmussen and Williams, 2006). Kriging can be briefly explained as follows.