Conformal e-prediction in the presence of confounding

TL;DR

This paper extends conformal e-prediction methods to handle confounding between observed data and labels, accommodating both IID and dependent data scenarios, enhancing predictive reliability in complex settings.

Contribution

It introduces a framework for conformal e-prediction that accounts for confounding and dependence, broadening its applicability beyond IID assumptions.

Findings

Method effectively adjusts for confounding effects.

Applicable to both IID and dependent data.

Improves prediction validity under complex data conditions.

Abstract

This note extends conformal e-prediction to cover the case where there is observed confounding between the random object $X$ and its label $Y$. We consider both the case where the observed data is IID and a case where some dependence between observations is permitted.