Calibration of stochastic computer simulators using likelihood emulation

Abstract

We calibrate a stochastic computer simulation model of ‘moderate’ computational expense. The simulator is an imperfect representation of reality, and we recognise this discrepancy to ensure a reliable calibration. The calibration model combines a Gaussian process emulator of the likelihood surface with importance sampling. Changing the discrepancy specification changes only the importance weights, which lets us investigate sensitivity to different discrepancy specifications at little computational cost. We present a case study of a natural history model that has been used to characterise UK bowel cancer incidence. Data sets and computer code are provided as supplementary material.

Metadata

Authors/Creators:

Oakley, J.E.
Youngman, B.D.

© 2016 The Author(s). Published by Taylor & Francis. This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. Permission is granted subject to the terms of the License under which the work was published. Please check the License conditions for the work which you wish to reuse. Full and appropriate attribution must be given. This permission does not cover any third party copyrighted material which may appear in the work requested.

Keywords:

Bayesian inference; computer experiments; natural history model; importance sampling; Gaussian process emulator

Dates: