I have a 3 dimensional mass spect dataset with x, y and z...z is the frequency. I am trying to apply PCA to this dataset. If size of matrix (M) is 10 X 10 X 2000 for x,y,z, how should I restructure this matrix ?? Should it be (10 X 10) X 2000 or 2000 X (10 X 10) for applying princomp correctly ?