principal components menu

This menu lets you perform principal components analysis on the objects you are analyzing. Principal components analysis is a technique
that generates those linear combinations of object measures (called eigenvectors) which express the greatest statistical variance over
all of the objects in your image. This analysis can sometimes be useful when there are hidden dependencies between different object
measures.
After you have performed a principal components analysis, the Create Graph menu will display additional options
under the Graph type menu field. If you set the Graph type to P.C. Highlight
or P.C. Constraint, the Axis 1 and Axis 2 fields will offer numbered principal
components instead of the usual object measures.
To perform a principal components analysis, use the following procedure:
- Set the Standardize? toggle to indicate whether you want the object measures to be standardized before the analysis is performed.
Standardization alters the object measures internally to make each measure have a mean of zero and a standard deviation of one. This
is usually a good idea, since it prevents one measure from predominating over another simply because of the units used to express each measure.
- Click on DO PCA. A list of the available object measures will appear. Click once on each measure you want to include in the analysis. Then
- Click on Exit. The system will proceed with the analysis and will display the eigenvectors and associated variances in a separate display on
the screen.