Computes point forecasts and prediction intervals for transfer function models.
# S3 method for tfm
predict(
object,
newdata = NULL,
y = NULL,
ori = NULL,
n.ahead = NULL,
level = 0.95,
i = NULL,
envir = NULL,
...
)Object of class predict.tfm containing:
Complete series including forecasts
Root mean square error for each forecast
Lower and upper prediction interval bounds (matrices)
Confidence level(s) used
Time points for all observations
Forecast origin (index and date)
Number of forecasts
A fitted tfm object.
Optional matrix or vector of future values for exogenous regressors and inputs. Rows correspond to forecast horizon, columns to predictors.
Optional ts object for alternative output series.
Forecast origin (observation index). Default is last observation.
Number of steps ahead to forecast. Default is series frequency.
Confidence level(s) for prediction intervals (0-1). Default is 0.95. Can be a vector for multiple intervals.
Optional differencing operator (lagpol) to apply before forecasting.
Environment for evaluation. NULL uses calling environment.
Additional arguments (currently unused).
Future values for transfer function inputs can be provided in three ways:
(1) extending input series beyond output length, (2) automatic forecasting
from associated um models, or (3) via the newdata argument.
If Box-Cox transformation was used, forecasts are back-transformed and intervals adjusted accordingly.
tfm, fit.tfm