Competition on Spatial Statistics for Large Datasets

As spatial datasets are becoming increasingly large and unwieldy, exact inference on models becomes computationally prohibitive. Various approximation methods have been proposed to reduce the computational burden. Although comprehensive reviews these exist, comparisons of their performances limited small medium sizes for a few selected methods. To achieve comparison comprising as many possible, we organized Competition Spatial Statistics Large Datasets. This competition had following novel features: (1) generated synthetic with ExaGeoStat software so that number realizations ranged from 100 thousand 1 million; (2) systematically designed data-generating represent processes wide range statistical properties both Gaussian non-Gaussian cases; (3) tasks included estimation prediction, results were assessed by multiple criteria; (4) made all publicly available serve benchmark other In this paper, disclose details along some analysis outcomes.