- Change default argument
optimControl = list(..., usegr = FALSE)
indensityMclustBounded()
andMclustBounded()
.
- Added
GMMlogreturn()
to model log-returns financial data via GMMs. - Added
VaR()
andES()
to compute risk measures from GMMs. densityMclustBounded()
allows to have lambda values fixed for a subset of variables while to estimate lambda parameters for the remaining variables.densityMclustBounded()
added noise component.densityMclustBounded()
andMclustBounded()
include an optional argumentcriterion
to specify the information criterion for model selection.- converted all documentation to use
roxygen2
package. - clean-up package dependencies.
- Starting with this version mclust >= 6.1 is needed because of using
efficient
softmax()
andlogsumexp()
functions coded in Fortran. This replaces previously included functions of the same name (now removed). - Added
prior
andnstart
arguments todensityMclustBounded()
function. - Added
rangepowerBackTransform()
. - Exported both
rangepowerTransform()
andrangepowerBackTransform()
. - Bug fixes.
- Bug fixes.
- Small bug fix in help page.
- Add code for computing entropy via Gaussian mixtures. See
help(EntropyGMM)
. - Add efficient Rcpp-based functions to compute log-sum-exp and softmax.
- Bug fixes.
- Allowed to have both
lbound = NULL
andubound = NULL
, in which case only power transformation is performed.
- Release submitted to CRAN.
- Add logarithm arg to
predict.densityMclustBounded()
function.
- Release for R version 4.0
- Inclusion of
MclustMEM()
and accompanying functions.
- Inclusion of
densityMclustBounded()
and accompanying functions.
- Initial developing release.