It seems that eulerr simply tries to minimize the residuals when laying out the plot. For plot with many areas of overlap, some of which can be small, eulerr thus doesn't plot some of these small areas of overlap, which can be problematic.
It might make sense to add a mode where plotting all areas of overlap is given priority over making all area sizes as close as possible to the data?
Here's an example:
original fitted residuals regionError
A 534 534 0 0.000
B 22 22 0 0.000
C 346 346 0 0.000
D 0 0 0 0.000
A&B 57 57 0 0.000
A&C 132 132 0 0.000
A&D 41 41 0 0.000
B&C 23 23 0 0.000
B&D 0 0 0 0.000
C&D 2 2 0 0.000
A&B&C 86 86 0 0.000
A&B&D 13 13 0 0.000
A&C&D 23 23 0 0.000
B&C&D 1 0 1 0.001
A&B&C&D 36 36 0 0.000
The B&C&D overlap is completely missing from the plot:
(A is Path1, B=2, C=3, D=Volunteers)
It seems that eulerr simply tries to minimize the residuals when laying out the plot. For plot with many areas of overlap, some of which can be small, eulerr thus doesn't plot some of these small areas of overlap, which can be problematic.
It might make sense to add a mode where plotting all areas of overlap is given priority over making all area sizes as close as possible to the data?
Here's an example:
The
B&C&Doverlap is completely missing from the plot:(A is Path1, B=2, C=3, D=Volunteers)