Hello:
I would need some help on the use of RVM in the kernlab package.
Questions:
1. Is there an optimal ratio between available number of cases and number of
vectors?
2. Does something like "overfitting" exist? If so how can be
detected/avoided?
3. The RVM model obtained is a S4 object. How can be saved and used in another
session? Can it be exported as a C or VisualBasic file?
Thank you very much
Zack Holmes
[[alternative HTML version deleted]]