gmatools

gmatools: Geometric multivariate analysis in R

TODO short description

See NEWS for the current version of gmatools and recent changes.

Installation

You can install the development version of gmatools from GitHub with:

# install.packages("devtools")
devtools::install_github("schtepf/GMA/pkg/gmatools")

Example

TODO show an illustrative example

library(gmatools)
## basic example code

Future work

• Package documentation

  • expand package documentation, adding code examples where possible
  • write a tutorial vignette based on data sets from corpora package

• Software tests

  • write software tests for various GMA functions (to the extent possible)

• Integrate further existing functions into gmatools

  • discriminant.plot() as gma.discriminant.plot()
  • gma.plot.features()
  • export .scaleMargins() as gma.scale.margins() or provide a dedicated method for feature contributions
  • correlation.plot() ?
  • parcoord.plot() ?
  • lda.cv() as gma.lda.cv(), but needs to be rethought with additional parameters
  • gma.clust(), possibly also utilities majorityLabels() and adjustedRandIndex()
  • interactive 3D visualisations (depending on rgl package)
    • gma.3d() and associated functions (need to add Suggest: rgl and conditional checks in function)
    • make.movie() is probably too specialised, interactive WebGL displays are better for end users

• Implement GMA$drop() method to remove basis dimensions

• Possibly implement rotations from factor analysis? (varimax(), packages GPArotation and psych)

  • use same interface as factanal() rotations, so an arbitrary function can be supplied by user (which means we must pass something that can be interpreted as a loadings matrix)
  • only orthogonal rotations will be allowed in GMA, of course