On the classification error rates in terms of semivariograms for Gaussian universal kriging models

Abstract

Bayes multiclass classification of spatial Gaussian data following the universal kriging model is considered. The closed-form expressions for the maximum likelihood (ML) estimator of regression parameters and the actual error rate (AER) in terms of semivariograms are derived.

Keywords

• pairwise discriminant function
• semivariogram
• actual error rate