Extends the capabilities for flexible partitioning and model-based clustering available in the packages 'flexclust' and 'flexmix' to handle ordinal and mixed-with-ordinal data types via new distance, centroid and driver functions that make various assumptions regarding ordinality. Using them within the flex-scheme allows for easy comparisons across methods.
This R package and github repository is an add-on package to the
R packages flexclust
and flexmix. These two root
packages are suites for flexible clustering with both partitioning
and model-based methods allowing for easy method variation and comparison.
With flexord, we provide new distance and centroid functions and new
model drivers for component distributions that are tailored towards
clustering ordinal or mixed-with-ordinal data.
We wrote this package to accompany the paper "Ordinal clustering in the flex-scheme" (2025)
by Ernst D, Ortega Menjivar L, Scharl T, Grün B, currently submitted to AJS.
In this paper, we reviewed methods available for ordinal data clustering and applied many
of the methods in our extensive simulation study. The replication code for the simulation study itself is collected
in the repository AJS-flexord.
You can find a more end user friendly introduction to the package, and a listing of the new methods we provide here.
The stable version of the package is available at CRAN, and can be installed via:
install.packages("flexord")
You can install the developer version of the package via:
devtools::install_github("dernst/flexord")