ABSTRACT
We show that several well-known statistical techniques have a similar structure. This structure can also be generalized to prediction problems that do not involve sampling and identical repetitions. We describe an axiomatization of the general prediction rule.
