Added Forecast() S4 class.
Added is_forc() function.
Added oos_realized_forc() function.
Added oos_lag_forc() function.
Added oos_vintage_forc() function.
Added conditional_forc() function.
Added historical_mean_forc() function.
Added random_walk_forc() function.
Added autoreg_forc() function.
Added mse_weighted_forc() function.
Changed the name of the collect() function to forc2df() to avoid namespace conflict with dplyr::collect().
Altered mse() and rmse() methods so that forecast accuracy can be calculated if there are NA forecast or realized values.
Altered autoreg_forc() so that AR models are properly computed using one to ar_lags number of lags and h_ahead forecasts are computed iteratively.
Added mae(), mape(), and R2() methods for evaluating forecast accuracy.
Altered estimation_end argument so that the origin of the first forecast is always greater than or equal to the estimation_end time.
Changed historical_mean_forc() to historical_average_forc() and altered the function so that forecasts can be calculated using either the historical mean or historical median. Also altered the function so that forecasts can be calculated if there are NA values in realized_vec.
Added return_betas argument to all applicable functions. If set to TRUE, returns a data frame of the coefficients used to create the forecast in each time period to the Global Environment.
Created str method for Forecast objects.
Added states_weighted_forc() function for computing state weighted forecasts.
Changed name of mse_weighted_forc() to performance_weighted_forc() to reflect that errors may be either MSE or RMSE.