General Suggestions

After you have examined the initial results, you will probably want to refine your analysis by changing some of the specifications for the nonlinear canonical correlation analysis. Here are some tips for structuring your analysis:

• Create as many sets as possible. Put an important variable that you want to predict in a separate set by itself.
• Put variables that you consider predictors together in a single set. If there are many predictors, try to partition them into several sets.
• Put each multiple nominal variable in a separate set by itself.
• If variables are highly correlated to each other, and you don’t want this relationship to dominate the solution, put those variables together in the same set.