Hello,
I have these kind of matrices (below) and I'd like to identify unique values specific to different groups of columns:
For instance, in the example above, if we decide that A, B, C are "Group 1" and D, E, F are "Group 2", and that the values in rows 1, 2, 3 are independent (i.e., "x" in "1" is not comparable to "x" in "2", etc):
- In condition (row) 1: "x" is a specific value only found in Group 1
- In condition 1: "y" is a specific value only found in Group 2
- In condition 2: "x" is a value found in majority in Group 1
- In condition 3: no specific value can be associated to Group 1 or 2.
What I would like to get is a measure of whether:
(1) there are values over-represented in one of the groups, or 100% specific to one group.
(2) what are these values
(3) if multiple values are a bit tricky, then: what is the value which is the most over-represented in one group compared to the other (the maximum being 100% in one group and 0% in the other)
Advice is very welcome, and if there could be a simple way to do this, that would be better. I have the feeling that the answer to this request could go beyond simple formulas (and therefore much outside my comfort zone!), but I'll try!
Thank you for your help
Best,
GM
Bookmarks