One can verify that, in canonical space, the among-group dispersion matrix A is equal to the matrix of eigenvalues and that the pooled within-groups dispersion matrix V is the identity matrix I. Beware: some computer programs calculate the discriminant scores as XU instead of [X centred] U or [X centred] C.

The classification functions, computed from eq. 11.41, are the following for descriptors xi and of the example:

Group 1: Scorei = -13.33333 + 8.00000 xi1 + 8.00000 xi2

Group 2: Scorei = -253.80000 + 36.80000 xi1 + 32.80000 xi2

Group 3: Scorei = -178.86667 + 34.93333 xi1 + 20.26667 xi2

The scores of the 7 objects i, computed from these functions, are the following:

Object number

Observed group

Function 1

Was this article helpful?

0 0

Post a comment