The authors address the lack of finite-sample analysis for prediction set sizes in conformal predictors, providing theoretical quantification and empirical estimation methods.
