Minor Changes : Added references to the accompanying
paper in the Journal of Statistical Software (doi:10.18637/jss.v117.i06), in the CITATION
file, the DESCRIPTION and the documentation.
Minor Changes : Deleted reference to a non-existing vignette.
Usage Changes : Added a palette
argument to Lorenz.graphs() and to the plot()
methods of "LR" and "PLR" objects.
Usage Changes:
Lorenz.curve no longer displays a graph but
only returns the function. Graph of Lorenz curves are obtained with
Lorenz.graphs.Lorenz.boot has a show_progress
argument to display a progress bar.store_LC=TRUE in function
Lorenz.boot.Lorenz.GA has a suggestions
argument to provide initial guesses to the genetic algorithm.Technical Changes:
Gini.coef.Usage Changes:
Lorenz.boot and PLR.CV no longer
have a data.orig argument. The original data are now
retrieved from the input LR or PLR
object.Breaking Changes:
Lorenz.Reg: The function structure has changed. It now
acts as a wrapper for the fitting functions Lorenz.GA,
Lorenz.FABSand Lorenz.SCADFABS. It returns an
object of class ''LR'' (non-penalized regression) or
''PLR'' (penalized regression). Each class has a set of
designated methods.Lorenz.boot: This function performs bootstrap
calculations for objects of class ''LR'' or
''PLR''. It uses the boot function from the
boot package. For penalized regression, it also computes an
out-of-bag score for tuning parameter selection. The function returns
the updated object with bootstrap results and adds the class
''LR_boot'' (non-penalized regression) or
''PLR_boot'' (penalized regression).PLR.CV: This function performs cross-validation for
objects of class ''PLR''. It computes a cross-validation
score for tuning parameter selection. The folds are constructed using
vfold_cv from the rsample package. The
function returns the updated object with cross-validation results and
adds the class ''PLR_cv''.''LR'' and
''PLR'' is now documented in the Lorenz.Reg
help page. Each method also has its own help page.New Features:
grid.arg and grid.value arguments in
Lorenz.Reg allow users to specify one tuning parameter for
penalized regression and construct a grid for it. Fitting is repeated
for each grid value, and optimal values are determined using available
methods (among BIC, bootstrap and cross-validation).diagnostic.PLR provides diagnostic information for
penalized Lorenz regression.fitted, explainedIneqand
autoplot were added for objects of classes
''LR'' and ''PLR''.Technical changes
Lorenz.GA (and the underlying
Rcpp functions) in order to ensure full reproducibility of results.seed argument.
This argument sets a local seed, used for the generation of random
objects within the function. the seed is reverted to its previous state
after the operation. This ensures that the seed settings do not
interfere with the global random state or other parts of the code.NEWS.md file to track changes to the
package.