PRIORS Subcommand (DISCRIMINANT command)

By default, DISCRIMINANT assumes equal prior probabilities for groups when classifying cases. You can provide different prior probabilities with the PRIORS subcommand.

  • Prior probabilities are used only during classification.
  • If you provide unequal prior probabilities, DISCRIMINANT adjusts the classification coefficients to reflect this.
  • If adjacent groups have the same prior probability, you can use the notation n*c on the value list to indicate that n adjacent groups have the same prior probability c.
  • You can specify a prior probability of 0. No cases are classified into such a group.
  • If the sum of the prior probabilities is not 1, the program rescales the probabilities to sum to 1 and issues a warning.

EQUAL. Equal prior probabilities. This is the default.

SIZE. Proportion of the cases analyzed that fall into each group. If 50% of the cases included in the analysis fall into the first group, 25% in the second, and 25% in the third, the prior probabilities are 0.5, 0.25, and 0.25, respectively. Group size is determined after cases with missing values for the predictor variables are deleted.

Value list. User-specified prior probabilities. The list of probabilities must sum to 1.0. The number of prior probabilities named or implied must equal the number of groups.

Example

DISCRIMINANT  GROUPS=TYPE(1,5)
  /VARIABLES=A TO H
  /PRIORS = 4*.15,.4.
  • The PRIORS subcommand establishes prior probabilities of 0.15 for the first four groups and 0.4 for the fifth group.