Select the analysis options for Discriminant Analysis

Stat > Multivariate > Discriminant Analysis > Options

In This Topic

Prior probabilities
Predict group membership for
Display of Results

Prior probabilities

If you know the prior probabilities of membership in the groups, enter the values. You can type in the probabilities or specify constants (K) that contain stored values. Enter one value for each group.

For example, if 60% of a population belongs to Group A and 40% belongs to Group B, the prior probabilities are 0.60 and 0.40.

Entering prior probabilities can improve the results of the analysis. Minitab assigns the first value to the group with the smallest (lowest) group identifier, the second value to the group with the second smallest (lowest) group identifier, and so on. If the group identifiers are text values, Minitab determines their rank by their value order. For more information on value order, go to Change the display order of text values in Minitab output.

Note

If the probabilities do not sum to 1, Minitab normalizes them. If you do not enter prior probabilities, Minitab uses equal probabilities for the analysis.

For more information on prior probabilities, go to What are posterior probabilities and prior probabilities?.

Predict group membership for

To classify observations with unknown groups, enter a column or columns that contain the new data for each predictor. Each row contains the measurements on one observation. The number of constants or columns must equal the number of predictors. For more information, go to Use discriminant analysis to predict group membership for new observations.

Display of Results

Do not display: Suppress all results, but store results if indicated.
Classification matrix: Display only the classification matrix and a statistics table.
Above plus ldf, distances, and misclassification summary: Display the classification matrix, the squared distance between group centers, the linear discriminant function, and a summary of misclassified observations.
Above plus mean, std. dev., and covariance summary: Display the classification matrix, the squared distance between group centers, the linear discriminant function (or the Generalized Squared Distance table), a summary of misclassified observations, means, standard deviations, and covariance matrices, for each group and pooled.
Above plus complete classification summary: Display the classification matrix, the squared distance between group centers, the linear discriminant function, a summary of misclassified observations, means, standard deviations, covariance matrices, for each group and pooled, and a summary of how all observations are classified. Minitab indicates misclassified observations with two asterisks beside the observation number.