A general statistical model for computer experiments with time series output

Manufacturing processes increasingly rely on computer experimentation as a substitute for costly physical experimentation. However, computer experimentation may not be very efficient because it often relies on computationally intensive simulation (or computer) models. To address this computational problem, this paper proposes a general statistical model as a computationally fast approximation for computer models with time series output. More precisely, the statistical models will be regression models with input-dependent design matrix and input-correlated errors. An example from the automotive industry will be used to illustrate the methodology.