Canonical Scores

 

The Canonical Scores tab contains the following elements:

Multivariate Correlations

These are Pearson correlations between the canonical score variables. The correlations are typically zero.

Multivariate Scatterplot Matrix

This plots the canonical scores against each other, colored by the dependent variables. Models that predict the categorical dependent variable well exhibit clusters corresponding to the classes that are well separated.

You can use these plots to identify points that are not predicted well and to see which dimensions are associated with classification performance.

See Scatterplot for more information.