Nonparametric Importance Sampling With Stochastic Computer Models

Abstract

This study presents an importance sampling method for assessing system performance using stochastic computer models. To relieve the computational burden while improving estimation accuracy, variance reduction techniques can be employed. Importance sampling is one of the most popular variance reduction techniques to improve simulation efficiency in rare event simulation. We consider problems where multiple input variables affect the simulation response and each input variable’s effect on the response depends on other factors. We devise a new nonparametric importance sampling method that can quantify the contributions of each input factor and its interactions with other factors, while avoiding computational problems and data sparsity issue arising in rare event simulation.