1) Solve the generalized eigenvalue problem: Reference from 114: - compute B , the covariance matrix of M (B for between-class covariance), and its eigen- decomposition B = V D B V T . The columns v of V in sequence from first to last define the coordinates of the optimal subspaces. Combining all these operations the th discriminant variable is given by Z = v T X with v = W 2 1 v Fisher arrived at this decomposition via a different route, without referring to Gaussian distributions at all. He posed the problem: Find the linear combination Z = a T X such that the betweenclass variance is maximized relative to the within- class variance..