=============================================================================================================== "Integrating uncertainty in time series population forecasts: An illustration using a simple projection model" Research Material Guy J. Abel*, Jakub Bijak, Jonathan J. Forster, James Raymer, Peter W.F. Smith and Jackie S.T. Wong * Address for correspondence: guy.abel@oeaw.ac.at =============================================================================================================== The supplementary zip archive for this paper contains a single R script file, cv_bma.R. The file can also be run from R directly having installed the tsbridge pacakge (i.e. you do not need to download the zip file) using: demo(cv_bama, package = "tsbridge", ask = FALSE) The cv_bma.R script illustrates how to estimates parameters and calculate posterior model probabilities for a set of constant variance models (with different lags) fitted to the full England and Wales data set. With a few minor adjustments this code can be adapted to estimate parameters for stochastic volatility and random variance shift models, also illustrated in the paper. However, the MCMC of these models will take a lot longer (especially the RV models), and hence are not demonstrated. The model averaging code can also be easily adapted to handle a larger model space, once parameters and normalising constants are estimated. Code should work for alternative data sets (such as shortened series). ===============================================================================================================