-
Within-class scatter defines the scatter of samples
around their respective means.
-
Between-class scatter defines scatter of expected
vectors around global mean.
-
Mixture-class scatter is the overall scatter obtained
from covariance of all samples
where
is the
overall mean.