Predict the next h steps ahead for the factor estimates and the data series. Given information up to time \(t\), a h-step ahead forecast is \(\bm{X}_{t+h}=\bm{\Lambda}\bm{A}^{h}\bm{F}_t+\bm{\Phi}^h\bm{\epsilon}_t\), where \(\bm{\Phi}=0\) for the IID idiosyncratic error case.
# S3 method for sparseDFM
predict(object, h = 1, standardize = FALSE, alpha_index = "best", ...)# S3 method for sparseDFM_forecast
print(x, ...)
X_hat \(h \times p\) numeric matrix of data series forecasts.
F_hat \(h \times r\) numeric matrix of factor forecasts.
e_hat \(h \times p\) numeric matrix of AR(1) idiosyncratic error forecasts if err=AR1 in sparseDFM.
h forecasts produced for h steps ahead.
err the type of idiosyncratic errors used in sparseDFM.
Prints out the h-step ahead forecast from predict.sparseDFM.
an object of class 'sparseDFM'.
integer. The number of steps ahead to compute the forecast for. Default is \(h=1\).
logical. Returns data series forecasts in the original data scale if set to FALSE. Default is FALSE.
Choose which L1 penalty parameter to display the results for. Default is 'best'. Otherwise, input a number between 1:length(alpha_grid) that indicates the required alpha parameter.
Further print arguments.
an object of class 'sparseDFM_forecast' from predict.sparseDFM.