Changes in version 2.2-0 (2026-01-29) User-visible changes o Made the package dependent on gamlss.dist. Minor changes: o Made sure that checks run without suggested packages. o Added vignette dependencies to vignette and removed hidden installation of package. o Updated URLs throughout the vignette. Changes in version 2.1-0 (2025-02-24) User-visible changes o Added new family DirichletLSS for Dirichlet regression models (implemented by Michael Balzer) Minor changes: o Improved error messages for families with censored response. Changes in version 2.0-7 (2023-03-09) Bug-fixes o Fixed minor bug of argument mismatch in internal S3 function get_qfun. o Updated inst/CITATION file to reflect current guidance on how to cite in packages. Changes in version 2.0-6 (2022-05-11) Bug-fixes o Fixed minor bug in cvrisk. o Fixed minor bug with out-of-bag risk for non-cyclical fitting. Changes in version 2.0-5 (2021-01-20) Miscellaneous o Added www and a / after the URLs to make CRAN happy. Changes in version 2.0-4 Miscellaneous o Changed http to https to make CRAN happy, hopefully finally. Changes in version 2.0-3 Miscellaneous o Minor changes to make CRAN happy. Changes in version 2.0-2 Bug-fixes o Initialize combined_risk in mboostLSS_fit to avoid writing (and overwriting) the combined risk in the global environment. Closes issue #49. o Fixed stabsel.mboostLSS for models fitted with FDboostLSS. Closes issue #51. o Code in man/families.Rd failed. Closes issue #53. o predint was not working when multiple variables were included within a single baselearner. Closes issue #55. o Fixed CRAN checks regarding all.equal by adding check.environment = FALSE. Changes in version 2.0-1 (2018-06-15) Miscellaneous o Export plot functions, streamlined interface of cvrisk and updated manuals. Closes issue #46. o Export predict.mboostLSS to make betaboost happy. o Updated reference for Thomas et al. Closes issues #47. Changes in version 2.0-0 (2017-05-05) User-visible changes o Added new fitting algorithm which can be used via the argument method: • method = "cyclic" is the standard approach which was also used previously. • method = "noncyclic" provides a new non-cyclical algorithm. The "noncyclic" method allow faster cross-validation and better results for stability selection (see below). For details see Thomas, J., Mayr, A., Bischl, B., Schmid, M., Smith, A., and Hofner, B. (2017), Gradient boosting for distributional regression - faster tuning and improved variable selection via noncyclical updates. _Statistics and Computing_. Online First. DOI 10.1007/s11222-017-9754-6 (Preliminary version: ). o Stability selection (stabsel) implemented for gamboostLSS models. o Models can now be fitted with zero steps (i.e., models containing only the offset). This change was mainly realized with the release of mboost 2.8-0. This closes issue #22. o The as.families interface to incorporate gamlss distributions for gamboostLSS models was adapted to allow for other link functions. o Added option stabilization = "L2" to use the average L2 norm of the negative gradient to make the updates for the different distribution parameters comparable. Bug-fixes o Fixed as.families("BB") and as.families("BI"), which is actually an mboost family. Closes issue #12. o Fixed as.families for 4-parametric "Mixed"-type families (e.g. as.families("BEOI") and as.families("BEINF")). Closes issue #28. Miscellaneous o Added Janek Thomas as new author. o Updated inst/CITATION due to release of JSS tutorial paper and added new noncyclical fitting paper (also to manuals). The latter closes issue #27. o Updated URL in DESCRIPTION. o Adapt mboostLSS_fit() such that it can be called by FDboost::FDboostLSS(). Changes in version 1.2-1 (2016-03-11) User-visible changes o Speed up computations by using response(fitted()) instead of fitted(, type = response) as the latter uses predict(). Closes issue #5. o Improved output of print.mboostLSS() and summary.mboostLSS(). Closes issue #7. o Added gamboostLSS_intern() to make FDboost happy. Caution: Do not use this function. Miscellaneous o Fixed inst/CITATION. Bug-fixes o Fixed tests for stabilization. o Fixed bug in cvrisk.mboostLSS() which occurred if families wasn't explicitly specified when fitting the model. Closes #9. Changes in version 1.2-0 (2015-08-19) User-visible changes o New interface: Stabilization of negative gradient is now specified via families, e.g., GaussianLSS(stabilization = "MAD"). Using options(gamboostLSS_stab_ngrad = TRUE) is still supported for backward compatibility but discouraged. o Added weighted.median which is used in "MAD" stabilization (see bugfixes). o Added preliminary version of the tutorial paper as vignette, which is now available as vignette("gamboostLSS_Tutorial", package = "gamboostLSS"). o Added summary function (request #2). Miscellaneous o Added preliminary version of tutorial paper as reference. Bug-fixes o Improve stored calls (closes #3). o Use weighted.median to compute MAD when negative gradients are stabilized. o Fixed handling of gamlss.dist families with type = "Mixed". o Fixed a bug in as.families with truncated families (which can be obtained using the package gamlss.tr). o The internal helper get_qfun is now a generic function. o Adhere to CRAN policies regarding import of base packages (closes #1). Changes in version 1.1-3 (2015-01-12) Miscellaneous o Changed plain text NEWS to inst/NEWS.Rd Bug-fixes o Changes in inst/CITATION to make CRAN happy: Citations can now be extracted without the need to install the package. o Corrected india data set: stunting ranged from -600 to 600 but should range from -6 to 6 o Fixed unstated dependencies o Changed require to requireNamespace and added gamlss.dist:: where necessary o Fixed bug in selected.mboostLSS o predict() returned a vector instead of a matrix o factors were coerced to characters in predint() Changes in version 1.1-2 (2014-06-26) Miscellaneous o added survival to suggests and load it in tests to reflect recent changes in mboost o removed the generic function risk() as this function is now defined in mboost Bug-fixes o check if response is correct for the family before the offset is computed Changes in version 1.1-1 (2014-06-17) Miscellaneous o re-added tests o changed \dontrun to \donttest in examples Changes in version 1.1-0 (2014-06-13) User-visible changes o added new function cvrisk for generic cross-validation (gamboostLSS now requires mboost >= 2.2-3 for the generic function definition) o added a function to obtain cross-validation grids (make_grid) o added new function as.families to include gamlss families (actually provided via gamlss.dist) in gamboostLSS o added new families BetaLSS (implemented by Florian Wickler), GammaLSS, GaussianLSS, ZIPoLSS, ZINBLSS o GaussianLSS is now the default for families o added new functions predint() and plot(predint()) to obtain and plot marginal prediction intervals o added stabilization factor for negative gradients (MAD): turn on the stabilization by using options(gamboostLSS_stab_ngrad = TRUE) o added new data sets india and india.bnd Miscellaneous o lots of minor improvements in the interface and methods o improved manuals in various places Bug-fixes o Speed of model fitting was dependent on iteration, i.e. the algorithm got slower for larger mstop values Changes in version 1.0-3 (2011-11-17) Miscellaneous o removed duplicated (and slightly altered) code from mboost that was required to make gamboostLSS work with earlier versions of mboost; gamboostLSS now requires mboost >= 2.0-12 Bug-fixes o fixed potential problem that might occur if users specify different outcomes for different components o fixed bug in tests/regtest-gamboostLSS.R Changes in version 1.0-2 (2011-08-31) Bug-fixes o changed the way we use environments (to get rid of note: no visible binding) Changes in version 1.0-1 (2011-08-12) Miscellaneous o updated dependencies Bug-fixes o fixed bug in selected() Changes in version 1.0 Miscellaneous o updated references o moved gamboostLSS to CRAN Bug-fixes o some changes to make gamboostLSS work with mboost 2.0.x series Changes in version 0.5 o initial development version of gamboostLSS on R-forge