A0519
Title: Conformal prediction for interval outcomes
Authors: Weiguang Liu - UCL (United Kingdom) [presenting]
Abstract: Interval censoring often occurs in economic data, leading to partial identification. Predicting outcomes that are interval censored is crucial in many practical scenarios. A technique is suggested to build prediction sets for such interval outcomes by leveraging a description of the precisely identified region of conditional cumulative distribution functions (CDF) combined with conformal inference. This approach ensures finite-sample coverage guarantees and achieves asymptotic efficiency.
