ABSTRACT
Discriminant analysis is all about what makes groups different from each other. You know that you have groups in your dataset but can you tell what group a case should belong to just from the variables you have? If you can, then you have variables that discriminate between the groups. But are all the variables useful for doing this, or just some? And how good are the variables at discriminating? Are they so great that they will always predict the correct group, or is there some uncertainty involved? Maybe for some cases you do not know the group to which they belong. Can information about cases whose group you do know help you allocate these other cases to groups?
