Conformal e-prediction

TL;DR

This paper re-examines conformal e-prediction, highlighting its advantages and relationships with conformal prediction, emphasizing its simplicity, design flexibility, and validity guarantees, offering a modern perspective on its utility.

Contribution

It systematically analyzes conformal e-prediction's properties, advantages, and its relation to conformal prediction, providing insights into its practical applications and theoretical foundations.

Findings

Conformal e-prediction is simpler and easier to design than conformal prediction.

Cross-conformal e-predictors have guaranteed validity, unlike their conformal counterparts.

Conformal e-prediction can emulate many advantages of conformal prediction.

Abstract

This paper discusses a counterpart of conformal prediction for e-values, conformal e-prediction. Conformal e-prediction is conceptually simpler and had been developed in the 1990s as a precursor of conformal prediction. When conformal prediction emerged as result of replacing e-values by p-values, it seemed to have important advantages over conformal e-prediction without obvious disadvantages. This paper re-examines relations between conformal prediction and conformal e-prediction systematically from a modern perspective. Conformal e-prediction has advantages of its own, such as the ease of designing conditional conformal e-predictors and the guaranteed validity of cross-conformal e-predictors (whereas for cross-conformal predictors validity is only an empirical fact and can be broken with excessive randomization). Even where conformal prediction has clear advantages, conformal e-prediction can often emulate those advantages, more or less successfully.