NEWS.md
In ineswilms/bigtime: Sparse Estimation of Large Time Series Models

bigtime 0.2.3

package documentation adjusted due to breaking change in roxygen2 7.0.0

bigtime 0.2.2

verbose arguments added to functions sparseVAR and sparseVARX
correction lambda grid to hvaralgorithms
removal SystemRequirements: C++11

bigtime 0.2.1

Dependency on Rcpp version >=1.0.7 added to the DESCRIPTION.

bigtime 0.2.0

Important starting note: Please update all R packages on which bigtime depends upon installing bigtime from CRAN. We have been informed on errors occurring when using older versions of the Rcpp package on which bigtime depends. In version 0.2.1 of bigtime (currently only available on github), we have therefore added the dependency on Rcpp version >=1.0.7
sparseVAR, sparseVARMA, sparseVARX are all more efficient and faster now
All estimation functions return corresponding S3 classes
fitted and residuals functions were implemented for VAR, VARMA, VARX
diagnostics_plot was implemented for VAR, VARMA, VARX
plot_cv was implemented; Allows to investigate the behavior in CV
sparseVAR, sparseVARMA and sparseVARX can now use information criteria to select the optimal penalization
simVAR allow for the simulation of VAR models using different sparsity patterns; utility functions summary and plot were also implemented for simulated VARs
Default selection procedure in sparseVAR, sparseVARMA and sparseVARX changed to "none". Will return a 3D array of estimations from this version on, unless a different selection procedure is chosen. WARNING: This can break old code!
recursiveforecast was implemented for VAR models.
is.stable was implemented for VAR models
in plot_cv, directforecast, and lagmatrix the model argument was removed. Functions know automatically what model was used. WARNING: This can break old code!
All model functions will give warnings if the given data is not standardized
added example data for VAR, VARMA, and VARX