RAFisher1

Discriminant analysis concerned with separating distinct sets of observations.

Discriminant Analysis is a multivariate technique concerned with separating distinct sets of observations to previously defined groups; rather exploratory in nature, it is a separatory procedure. Here, we develop this technique considering normal populations with equal covariance matrices. So, prior to the Discriminant Analysis it performs a Multivariate Analysis of Variance to test any difference between the group's means vector. Then it describe the differential features of the observations from the known populations by trying to find discriminants whose numerical values are such that the populations are separated as much as possible.

It needs to input the multivariate data matrix, vector of prior probabilities (unknown them,pp = 1; known them,pp = 2 and [you must to give it]) and significance level (default = 0.05).

As output it gives a complete Multivariate Analysis of Variance table, mean vectors, vector of prior probabilities, testing of the Mahalanobis distances among groups, classification functions (grouping), list of misclassifications and the total percent of correct classification.