Hi, Are there any methods for computing the correlation between two multi-dimensional matrices ? Will transforming the matrices into vectors and applying pearson be fine ? Any blind spots that I should be aware ? Thanks Stanley [[alternative HTML version deleted]]
Ng Stanley <stanleyngkl <at> gmail.com> writes:> > Hi, > > Are there any methods for computing the correlation between two > multi-dimensional matrices ? Will transforming the matrices into vectors and > applying pearson be fine ? Any blind spots that I should be aware ? >If they were something like multi-dimensional correlation matrices (i.e., the elements weren't all independent) then you might want to do something like implementing a multidimensional Mantel test ... Ben Bolker