I've got one dependent variable and two independent variables. I need to calculate the multiple regression function via canonical correlation analysis.

How can I solve this problem using the eigenstructure examination? I've one dependent so there are only within-correlations and no between-correlations which are necessary for finding the eigenvalues and eigenstructures, using singular value decomposition?