Regressions for Making Extrapolations

Abstract

Kriging, otherwise called cumulative Gaussian regression with an exponential model, is a statistical model where observations occur in a continuous domain, e.g., time or space. It uses matrix algebra to fit correlations between known and unknown places in time or space. A second methodology for making predictions about unmeasured places from measured ones is Markov regressions, just like kriging an exponential methodology, where, also with the help of matrix algebra, long term predictions can be made about short term observations. The current chapter reviews the two methods for extrapolations, and uses real and hypothesized data examples for the purpose.