The Capital Spectator’s combination forecasts are based on the following models:

Exponential smoothing state space model: the average forecast is used from 100 simulations based on bootstrap aggregating via the forecasting package. The data set is the historical record for the target indicator.

Autoregressive integrated moving average model: the average forecast is used from 100 simulations based on bootstrap aggregating via the forecasting package. The data set is the historical record for the target indicator.

Neural network model: the average forecast is used from 100 simulations via the forecasting package. The data set is the historical record for the target indicator.

Naïve model: this forecast simply extracts the last data point and assumes that it will prevail for the next 12 months.

Facebook’s Prophet forecasting tool. The data set is the historical record for the target indicator.

Theta method forecast model: the methodology is a simple exponential smoothing with drift via the forecasting package.

Bayesian Structural Time Series: Time series regression using dynamic linear models fit using a Markov chain Monte Carlo methodology via the bsts package, which was written by Google’s Steven Scott and Hal Varian.

Vector autoregression model: this multivariate methodology (via the vars package) uses several data sets to forecast the target indicator.