NEWS.md
In LorenzRegression: Lorenz and Penalized Lorenz Regressions

LorenzRegression 2.2.0

Usage Changes:

Function Lorenz.curve no longer displays a graph but only returns the function. Graph of Lorenz curves are obtained with Lorenz.graphs.
Function Lorenz.boot has a show_progress argument to display a progress bar.
Confidence bands can be obtained for the explained Lorenz curve. This requires setting store_LC=TRUE in function Lorenz.boot.
Function Lorenz.GA has a suggestions argument to provide initial guesses to the genetic algorithm.

Technical Changes:

Usage Changes:

Functions 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:

The package structure has been standardized and builds upon existing libraries. Please review the updated function documentation for current usage.
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''.
Method availability for classes ''LR'' and ''PLR'' is now documented in the Lorenz.Reg help page. Each method also has its own help page.

New Features:

The 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.
Methods fitted, explainedIneqand autoplot were added for objects of classes ''LR'' and ''PLR''.

Technical changes

Modifications occurred in Lorenz.GA (and the underlying Rcpp functions) in order to ensure full reproducibility of results.
Functions implying randomness have a 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.