Multivariate Methods > Discriminant Analysis
Publication date: 08/13/2020

Discriminant Analysis

Predict Classifications Based on Continuous Variables

Discriminant analysis predicts membership in a group or category based on observed values of several continuous variables. Specifically, discriminant analysis predicts a classification (X) variable (categorical) based on known continuous responses (Y). The data for a discriminant analysis consist of a sample of observations with known group membership together with their values on the continuous variables.

For example, you might attempt to classify loan applicants into three loan categories (X) based on expected profitability: low interest rate loan, long term loan, or no loan. You might use continuous variables such as current salary, years in current job, age, and debt burden, (Ys) to predict an individual’s most profitable loan category. You could build a predictive model to classify an individual into a loan category using discriminant analysis.

Features of the Discriminant platform include the following:

A stepwise selection option to help choose variables that discriminate well.

A choice of fitting methods: Linear, Quadratic, Regularized, and Wide Linear.

A canonical plot and a misclassification summary.

Discriminant scores and squared distances to each group.

Options to save prediction distances and probabilities to the data table.

Figure 5.1 Canonical Plot 


Overview of the Discriminant Analysis Platform

Example of Discriminant Analysis

Launch the Discriminant Analysis Platform

Stepwise Variable Selection
Discriminant Methods
Shrink Covariances

Discriminant Analysis Report

Principal Components
Canonical Plot and Canonical Structure
Discriminant Scores
Score Summaries

Discriminant Analysis Options

Score Options
Canonical Options
Example of a Canonical 3D Plot
Specify Priors
Consider New Levels
Save Discrim Matrices
Scatterplot Matrix

Validation in JMP and JMP Pro

Statistical Details for the Discriminant Analysis Platform

Description of the Wide Linear Algorithm
Saved Formulas
Multivariate Tests
Approximate F-Tests
Between Groups Covariance Matrix
Want more information? Have questions? Get answers in the JMP User Community (