Properties and Comparison of Some Kriging Sub-model Aggregation Methods

Kriging is a widely employed technique across computer experiments, machine learning and geostatistics. An important challenge for kriging its high costs when dealing with large datasets. This article focuses on class of methods aiming at decreasing this computational burden by aggregating predictors based smaller data subsets. More precisely, it shows that aggregation ignore the covariance between sub-models typically yield inconsistent predictions, whereas nested method enjoys several attractive properties: consistent, can be interpreted as an exact conditional distribution modified prior, covariances given observations computed efficiently. also includes theoretical numerical analysis how assignment observation points to affect prediction ability aggregated model. Finally, extended measurement errors universal kriging.