Generate a representative powered fitting equation of data after using PCA

With a mxn data set (X), I've ran [COEFF,SCORE,latent,tsquare]= princomp(X). I'm attempting to represent one element of the data in terms of the others by way of a power method (e.g, e=W*f^X*g^Y*h^Z), where e, f, g and h are the data variables; and W, X, Y, Z are constant factors. In other words, I'm trying to determine what W,X,Y,Z should be after running princomp.

Thanks in advance for any assistance.

Regards, Paul