mu-comparison with many groups #102
Description
When summarizing more than 10 groups (e.g. an entire SSA) the graphical summaries of marginal distributions aren't feasible. It would be handy to graphically depict group-wise similarity via median values of each continuous variable, and maybe even factor variables via Gower's distance metric.
Something like this:
Groups could be evaluated according to one / more of:
- manual generalization of legend
- unsupervised classification
- tabular data summaries from component / muaggatt data
Output may be cluttered, so a more advanced labeling engine will be required. Also, searchable output would make the figure more far more useful. Consider SVG or {plotly} output devices.