This paper deals with the prediction of time series with correlated errors at each time point using a Bayesian forecast approach based on the multivariate Holt–Winters model. Assuming that each of the univariate time series comes from the univariate Holt–Winters model, all of them sharing a...