The Evaluate() method for Ranking takes a label and groupId as input, but unlike other Evaluators and the ranking trainers, it does not have default values, so these must always be specified. This leads to an inconsistent feeling across the API surface.