Work

Dynamic Cumulative Residual Fisher Information Measures, Connections to Aging Measures, and Use in Model Selection

Year: 2026

Type: article

Abstract: This paper proposes dynamic versions of the cumulative residual Fisher information and the relative cumulative residual Fisher information measures for residual lifetime variables. In this context, we provide representations of these proposed dynamic information measures in terms of hazard rate and mean residual lifetime functions. We also establish several theoretical results associated with these measures for two well-known models: equilibrium survival and proportional hazard functions. Further, we develop a stochastic ordering result for comparing the dynamic cumulative residual Fisher information for two residual lifetime variables. We then present nonparametric estimation methods for these dynamic information measures. The performance of the proposed estimators is evaluated using both simulated data and a real data example concerning failure times from repairable items. Finally, we show that the dynamic relative cumulative residual Fisher information can act as an effective model selection criterion, providing an alternative to some well-known measures.

Source: IEEE Transactions on Information Theory

DOI: 10.1109/tit.2026.3727364

Authors: Omid Kharazmi, Maryam Sharafi, Narayanaswamy Balakrishnan

Institutions: Vali Asr University of Rafsanjan, Razi University, McMaster University

Language: English

Topic: Statistical Distribution Estimation and Applications

Subfield: Statistics and Probability

Field: Mathematics

Domain: Physical Sciences

Open Access status: hybrid

Estimated APC paid: $2,645

OpenAlex ID: W7204185433

FWCI:

Cites: 24

Cited by:

Related to: 10