When performing the multivariates analysis called discriminant analysis in JMP, the score Summaries give a number of "Misclassified observations". The user manual pdf "multivariate methods" define the following:
Number Misclassified: Provides the number of observations in the specified set that are incorrectly classified. Percent Misclassified: Provides the percent of observations in the specified set that are incorrectly classified.
Prob(Pred): Estimated probability of the observation’s predicted classification.
It is unclear to me on what basis anobservation is well classified or not. In other words, which this the true criteria behind the "misclassified"?
1) Maybe it is Prob(Pred), but again, how is it calculated?
For instance, is it calculated from the following definitions (given in the "saved formula" section)
1) qt prior probability of membership for group t
2) the p(t|y) posterior probability that y belongs to group t is defined.
Discriminant analysis is a form of supervised learning. You provide a known response with the training data set. The discriminant analysis results in a predictive model. The predicted response is compared to the given response. If they don't match, then that observation is considered misclassified.