NEWS.md
In mvtboost: Tree Boosting for Multivariate Outcomes

mvtboost 0.4.4

mvtb.covex now estimates the covariance explained matrix, it is not estimated by default
Arguments in mvtb correspond directly to gbm
Tests are more flexible, maintainable, have improved coverage
Documentation for several functions has been simplified
mvtb.fit is now exported
Any distribution can be any supported by mvtb, but covariance explained can only be estimated when distribution=gaussian

Added mvtb.covex that directly returns the covariance explained matrix.
mvtb.plot gains return.grid argument to return gbm grid that is used for plotting
Documentation of functions improved, using more \code{} statements, among other things
Tests for summary functions improved
Revised package vignettes
corrected order of variable names for 'age', 'educ' and 'income' in wellbeing data
mvtb.heat gains a dec argument to specify the number of digits after the decimal to plot
example in ?mvtboost has been updated to produce a cleaner plot

Added wellbeing data and vignette
Improved documentation for mvtb, and other functions
mvtb.summary gains a covex argument if covex is not wanted
predict now drops unnecessary dimensions
cluster.covex becomes mvtb.cluster, a more generic way to cluster any table. obtains a 'plot' argument, a call to mvtb.heat
heat.covex becomes mvtb.heat, a generic way to heatmap any table (e.g. covex or relative influence). Has better plotting defaults to allow room for y-axis labels. Gets cexRow,cexCol arguments, with sensible defaults. Clustering can also be turned off by setting clust.method=NULL.
mvtb.perspec plots variable names by default, gets ylab, xlab, zlab arguments
fixed weighting influence bug in mvtb