Predicting the value of a variable at an unsampled location is presented for continuous variables, such as the grade of an element.
In this paper, the methods for estimation (prediction) are reviewed based on pure geometrical conditions, and then move on to introduce kriging, which is a collection of methods based on principles of regression.
These methods make use of the spatial continuity of the variables by means of the variogram or the covariance. We show simple kriging, which assumes the mean of the random variables is constant and known over the domain.
Author: Julián M. Ortiz
Download link below