Feature Idea: Alluvial Diagrams for McNemar instead of bar graphs

[sbalci]

Describe the solution you'd like
Alluvial Diagrams for McNemar instead of bar graphs.
I use repeated measures of categorical variables in my research. A marker positive in tumor may be positive or negative in metastasis. In alluvial diagrams I like to see the change as a flow. I think it is visually more informative.

How could we do it?
similarly alluvial, ggalluvial packages

[IndrajeetPatil]

Closing in favor of #500