Package: boostmtree Version: 1.5.1 BUILD: bld20220309 --------------------------------------------------------------------------------- CHANGES TO RELEASE 1.5.1 --------------------------------------------------------------------------------- Added the ability to model a nominal or ordinal longitudinal response. With this addition, package now handles continuous, binary, nominal or ordinal longitudinal response. RELEASE 1.4.1 --------------------------------------------------------------------------------- Added the ability to model a binary response. Added a separate function, vimp.boostmtree(), for calculating VIMP using out of bag or test data. This function also allows estimatation of separate as well as joint VIMP (of multiple covariates). RELEASE 1.3.0 --------------------------------------------------------------------------------- Fix to avoid mclapply on Windows, and to default to lapply instead. The in-sample CV now only uses out of bag data rather than the previous version which was using n-1 observations. Variable importance (VIMP) is calculated using the out of bag data in the grow mode. The name of the function vimp.plot has been replaced with vimpPlot. NULL values encountered in the mclapply function have been fixed. The marginalPlot function is new. It is similar to the partial plot except that it provide unadjusted predicted y values. RELEASE 1.2.1 --------------------------------------------------------------------------------- Fix to avoid mclapply on Debian, and to default to lapply instead. RELEASE 1.2.0 --------------------------------------------------------------------------------- Synchronization with package dependency in randomForestSRC (1.5.0) due to minor change in membership output protocols. RELEASE 1.1.0 --------------------------------------------------------------------------------- This update includes an "in sample" cross-validation method for determining the optimal boosting iteration. See the help file of boostmtree for more details. Various bugs and fixes are also addressed. RELEASE 1.0.0 --------------------------------------------------------------------------------- Please report bugs to ubk@kogalur.com