optimus - Model Based Diagnostics for Multivariate Cluster Analysis
Assessment and diagnostics for comparing competing
clustering solutions, using predictive models. The main
intended use is for comparing clustering/classification
solutions of ecological data (e.g. presence/absence, counts,
ordinal scores) to 1) find an optimal partitioning solution, 2)
identify characteristic species and 3) refine a classification
by merging clusters that increase predictive performance.
However, in a more general sense, this package can do the above
for any set of clustering solutions for i observations of j
variables.
