Title :
Confidence Intervals for Quantiles When Applying Latin Hypercube Sampling
Author :
Nakayama, Marvin K.
Author_Institution :
Comput. Sci. Dept., New Jersey Inst. of Technol., Newark, NJ, USA
Abstract :
Latin hypercube sampling (LHS) is a variance-reduction technique (VRT) that can be thought of as an extension of stratified sampling in higher dimensions. It can also be considered a generalization of antithetic variates, another VRT. This paper develops asymptotically valid confidence intervals for quantiles that are estimated via simulation using LHS.
Keywords :
covariance analysis; sampling methods; Latin hypercube sampling; confidence intervals; quantiles; simulation; stratified sampling; variance-reduction technique; Analytical models; Estimation; Hypercubes; Monte Carlo methods; Portfolios; Random variables; Stochastic processes; Latin hypercube sampling; confidence interval; quantile; variance reduction;
Conference_Titel :
Advances in System Simulation (SIMUL), 2010 Second International Conference on
Conference_Location :
Nice
Print_ISBN :
978-1-4244-7783-8
Electronic_ISBN :
978-0-7695-4142-6
DOI :
10.1109/SIMUL.2010.10
