EBMAforecast: Ensemble Bayesian Model Averaging Forecasts

No single model dominates across all forecasting tasks. EBMAforecast combines predictions from multiple component models using ensemble Bayesian model averaging (EBMA), weighting each model by its historical accuracy to produce superior out-of-sample forecasts. The package supports both expectation-maximization and fully Bayesian estimation via Gibbs sampling. Originally developed for political forecasting applications, it is broadly applicable wherever researchers wish to combine model-based predictions. Methodology described in Montgomery, Hollenbach, and Ward (2012), Political Analysis, and Montgomery, Hollenbach, and Ward (2015), International Journal of Forecasting.