Welcome to the componentts package, a package whose goal is to provide
ways of selecting the number of components to keep in a two-table (the
tt in componentts) analysis. The two-table analyses covered by this
package are partial least squares correlation (PLSC), canonical
correlation analysis (CCA), and redundancy analysis (RDA). componentts
is the result of my (Luke Moraglia’s) dissertation, where I investigated
several “stopping rules” for the two-table analyses—rules or algorithms
that select the number of components to keep. The stopping rules covered
by this package all involve permutation tests, from the very simplest
permutation tests, such as SRP(), to more complex rules with
additional cross-validation, such as multiple_hold().
For an overview of the functionality of the package, I recommend reading the accompanying vignette, and if you’d really like to become an expert, I recommend reading my dissertation, which will be available publicly once it has been defended (Summer 2024) and in the mean time is available by contacting me.
You can install componentts from GitHub with:
# install.packages("devtools")
devtools::install_github("LukeMoraglia/componentts")You may first need to install the GSVD package by Derek Beaton
(https://github.com/derekbeaton/GSVD). At time of writing, April 2024,
componentts relies on a bug fix to the GSVD package that allows
eigendecomposition on rank-one matrices. This bug fix has been
implemented in the master branch on my GitHub (the relevant issue is
#32 on Derek’s GitHub, and we are currently waiting on pull request
#33 to get this fix integrated into the official package).
devtools::install_github("LukeMoraglia/GSVD")