I wonder if anyone has a suggestion on how to solve the following problem.
I have a matrix of size n*p*t. I would like to find areas in the plane n*p, that co-vary (in the t dimension). In other words, sub-regions on the plane (n*p) whose values vary together along the t dimension.
An example:
` t=1
[ 0 0 0
[ 0 1 1
[ 0 1 1
--->
t=2
[ 3 2 7
[ 0 2 2
[ 4 2 2 `
So the bottom corner would be assigned to a group since the values co-vary.
Do you have any ideas how to tackle such a problem?
Thanks!
