Pre-Recorded Talk
in
Workshop: Workshop on Distribution-Free Uncertainty Quantification

Multi Split Conformal Prediction (Spotlight #10)

Aldo Solari · Vera Djordjilovic

Abstract:

Split conformal prediction is a computationally eā€€cient method for performing distribution-free predictive inference in regression. It involves, however, a one-time random split of the data, and the result can strongly depend on the particular split. To address this problem, we propose multi split conformal prediction, a simple method based on Markov's inequality to aggregate split conformal prediction intervals across multiple splits.