An object class to be used with cv.BigVAR
Dataa \(T \times k\) multivariate time series
model_dataprocessed time series and lag matrix
lagmaxMaximal lag order for modeled series
interceptIndicator as to whether an intercept should be included
StructurePenalty Structure
RelaxedIndicator for relaxed VAR
GranularityGranularity of penalty grid
horizonDesired Forecast Horizon
crossvalCross-Validation Procedure
MinnesotaMinnesota Prior Indicator
verboseIndicator for Verbose output
datesdates extracted from an xts object
icIndicator for including AIC and BIC benchmarks
VARXVARX Model Specifications
VARXIVARX Indicator
T1Index of time series in which to start cross validation
T2Index of times series in which to start forecast evaluation
ONESEIndicator for 'One Standard Error Heuristic'
ownlambdasIndicator for user-supplied lambdas
tfIndicator for transfer function
alphaGrid of candidate alpha values (applies only to Sparse VARX-L and Elastic Net models)
recursiveIndicator as to whether recursive multi-step forecasts are used (applies only to multiple horizon VAR models)
constvecvector indicating variables to shrink toward a random walk instead of toward zero (valid only if Minnesota is TRUE)
toloptimization tolerance
window.sizesize of rolling window. If set to NULL an expanding window will be used.
separate_lambdasindicator to use separate penalty parameter for each time series (default FALSE)
lossLoss function to select penalty parameter (one of 'L1','L2','Huber').
deltadelta parameter for Huber loss (default 2.5)
gammagamma parameter for SCAD or MCP penalty (default 3)
rolling_oosTrue or False: indicator to update the penalty parameter over the evaluation period (default False)
linearindicator for linearly decrementing penalty grid (FALSE is log-linear).
refit_fractionfraction of least squares refit to incorporate (default is 1).
To construct an object of class BigVAR, use the function constructModel
constructModel