Movatterモバイル変換


[0]ホーム

URL:


Build StatusCRAN versionCRAN RStudio mirror downloadsTotal CRAN downloadsCoverage Status

GMCM

Fastestimation of Gaussian Mixture Copula Models

TheGMCMpackage (Bilgrau et. al.,2016) offers R functions that perform high-dimensional meta-analysis(Li et. al., 2011) andgeneral unsupervised cluster analysis(Tewariet. al., 2011) using Gaussian Copula Mixture Models in a very fastmanner. Online documentation is availablehere.

Gaussian copula mixture models (GMCMs) are a very flexiblealternative to Gaussian mixture models in unsupervised cluster analysisfor continuous data where non-Gaussian clusters are present. GMCMs modelthe ranks of the observed data and are thus invariant to monotoneincreasing transformations of the data, i.e. they are semi-parametricand only the ordering of the data is important. Alternatively, aspecial-case of the GMCMs can be used for a novel meta-analysis approachin high-dimensional settings. In this context, the model tries tocluster results which agree and do not agree on statistical evidenceinto a reproducible and irreproducible group.

The optimization of the complicated likelihood function is difficult,however.GMCM utilizesRcpp andRcppArmadilloto evaluate the likelihood function quickly and arrive at a parameterestimate using either standard numerical optimization routines or anpseudo EM algorithm.

Additional information, documentation, help, and examples can befound byhere or byrunning?GMCM inR. The paper [1] is alsofound as a vignette byvignette("GMCM-JStatSoft"). The coreuser functions ofGMCM arefit.full.GMCMandfit.meta.GMCM.

Installation

The released and tested version ofGMCM is availableatCRAN(Comprehensive R Archive Network). It can be installed from within R byrunning

install.packages("GMCM")

If you wish to install the latest version ofGMCMdirectly from the master branch at GitHub, run

#install.packages("remotes")  # Install remotes if neededremotes::install_github("AEBilgrau/GMCM")

Note, that this version is in development and is likely differentfrom the version at CRAN. As such, it may be unstable. Be sure that youhave thepackagedevelopment prerequisites if you wish to install the package fromthe source.

When installed, runGMCM::runGMCM() to launch a localinstance of the GMCM shiny application also availableonline at shinyapps.io. Runnews(package = "GMCM") to view the latest changes of GMCMor visithere.

For previous versions ofGMCM, visit the oldreleases at GitHubor thearchive atCRAN.

References

  1. Anders Ellern Bilgrau, Poul Svante Eriksen, Jakob GulddahlRasmussen, Hans Erik Johnsen, Karen Dybkaer, Martin Boegsted (2016).GMCM: Unsupervised Clustering and Meta-Analysis Using GaussianMixture Copula Models. Journal of Statistical Software, 70(2),1-23.doi:10.18637/jss.v070.i02


[8]ページ先頭

©2009-2025 Movatter.jp