Hmm.. this is a good one. In short, I think with the dataset as is, Sets won't be of use because you can't base them on Table calculations. Likewise, it gets tricky because sorting and filtering on Table calcs is fiddly or not possible in the ways needed.
I've had a go with table calcs to get part way there - I suspect the grouping in to Other is doable.
Further thoughts are leading me to the other way of doing this sort of thing: duplicate the datasource and blend against the other year. Then everything's in one row and much easier to do. Sets would be possible then too I think.
Anyway, let this be the start of the conversation
I have come to realize that calculations cannot be used to create sets. I have tried a variety of different approaches using the posts below and your suggestions above as a guide, but have not had success.
I believe that I would need to use calculations no matter what to calculate variance, so it appears that this type of grouping cannot be done. Perhaps brighter minds on here can prove me wrong!
I have been able to created a sample of the type of dashboard I am looking for and I wanted to share the results here in case anyone else was looking to achieve the same results. I used the superstore data set for this model.
I have put some notes in the caption of a few tabs that outline some of the hurdles I had and the solution to each issue. Many thanks to the Tableau Support group for their help with this!