Estimating the variance of the predictor in stochastic Kriging
Kleijnen,J.P.C. Mehdad,Ehsan
Kleijnen,J.P.C.
Mehdad,Ehsan
Abstract
We study the estimation of the true variance of the predictor in stochastic Kriging (SK). First, we obtain macroreplications for a SK metamodel that approximates a single-server simulation model; these macroreplications give independently and identically distributed predictions. This simulation may use common random numbers (CRN). From these macroreplications we conclude that the usual plug-in estimator of the variance significantly underestimates the true variance. Because macroreplications of practical simulation models are computationally expensive, we next formulate two bootstrap methods that use a single macroreplication: (i) a distribution-free method that resamples simulation replications (within the single macroreplication), and (ii) a parametric method that assumes a Gaussian distribution for the SK predictor, and estimates the (hyper)parameters of that distribution from the single macroreplication. Altogether we recommend distribution-free bootstrapping for the estimation of the SK predictor variance in practical simulation experiments.
Description
Date
2016-08
Journal Title
Journal ISSN
Volume Title
Publisher
Research Projects
Organizational Units
Journal Issue
Keywords
Kriging, Gaussian process, predictor variance, plug-in, Bootstrap
Citation
Kleijnen, J P C & Mehdad, E 2016, 'Estimating the variance of the predictor in stochastic Kriging', Simulation Modelling Practice and Theory, vol. 66, pp. 166-173. https://doi.org/10.1016/j.simpat.2016.03.008
License
info:eu-repo/semantics/closedAccess