I don't think that there is a very strong reason for this lack. The ALR does behave very oddly in some respects relative to the normal model of an AVC, but as you say the classify* methods could easily delegate to the best performer.
On Thu, Nov 10, 2011 at 10:55 AM, Koert Kuipers <[email protected]> wrote: > Why doesn't AdaptiveLogisticRegression extend (or implement since it is > almost an interface) AbstractVectorClassifier? It could pass the classify > requests to it's best CrossFoldLearner. >
