Assessing Model Validity Using KL Divergence and Beta-Stacy Process for Right Censored Data

Abstract

Model validation constitutes a crucial component of statistical analysis, with heightened challenges encountered when addressing censored data. This paper introduces a novel approach that integrates the Kullback-Leibler divergence, the beta-Stacy process, and relative belief ratio inferences to form a comprehensive method for model validation. Significantly, this method is applicable to both complete and right-censored data. Theoretical results related to this approach are presented, accompanied by a computational algorithm for implementation. The effectiveness of the approach is illustrated through a series of data analysis examples, highlighting its good performance. MSC 2020 62N03, 62F15.