Description
Direct MLE for Multivariate Normal Mixture Distributions.
Description
Multivariate Normal (i.e. Gaussian) Mixture Models (S3) Classes. Fitting models to data using 'MLE' (maximum likelihood estimation) for multivariate normal mixtures via smart parametrization using the 'LDL' (Cholesky) decomposition, see McLachlan and Peel (2000, ISBN:9780471006268), Celeux and Govaert (1995) <doi:10.1016/0031-3203(94)00125-6>.
README.md
norMmix package
This package implements a model fitting algorithm for normal mixture models.
TODO:
- [X] make init method optional with no default
- [ ] spurious cluster suppression
- [X] (WIP) add vignette
- [X] test against windows
- [X] (WIP) reintroduce functionality for many MLE passes for model selection
Install
On linux machines, using R only:
git clone https://github.com/TrN000/norMmix.git
R CMD build norMmix
R CMD INSTALL norMmix_0.0-2.tar.gz
To build and check the package the following packages are necessary.
> install.packages(
c("mvtnorm", "mclust", "sfsmisc", "nor1mix", "testthat",
"knitr", "rmarkdown", "compositions", "mixtools"))