A geometric–exponential mixture model for joint inference of discrete and continuous lifetime data

Lifetime data from reliability studies are commonly observed continuously over time, but the observations may also be recorded discretely in case of periodic inspections. We study joint statistical inference based on independent lifetime tests with either continuously or discretely observed lifetime data. When the lifetimes of tested items are assumed to be exponentially distributed and if these are only inspected at non-negative integer time points, then the observations correspond to realizations of a discretized version of the lifetime distribution, namely to a geometric distribution with the same distribution parameter involved. We propose and examine a geometric–exponential mixture (GEM) model for the situation, where continuously recorded lifetimes supposed as observations from an exponential distribution as well as lifetimes from identically distributed items monitored at integer time points are available. Statistical inference is then based on all observations. What is most important with the GEM model is that mixtures of a continuous distribution and its discretized version having linked parameters constitute a regular exponential family such that well-known tools and statistical methodology can be utilized. From a minimal canonical representation of the exponential family associated with the GEM model, we obtain the cumulant function, the mean value function, and expressions for divergence measures. We then discuss the existence and uniqueness of maximum likelihood estimators for the distribution parameter and the mixture parameter; moreover, respective uniformly minimum-variance unbiased estimators are considered. Uniformly most powerful (unbiased) tests and uniformly most accurate (unbiased) confidence intervals can be derived. The proposed methods are illustrated by means of two simulation-based data examples. An outlook on a respective family of mixture distributions based on gamma distributions is given.

Authors

Institutions

Publication Details

Journal
Statistics
Published
2026-11-17
DOI
https://doi.org/10.1080/02331888.2026.2762388
Primary Topic
Statistical Distribution Estimation and Applications
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

A geometric–exponential mixture model for joint inference of discrete and continuous lifetime data

Udo Kamps, Zhe Su
Statistics
Statistical Distribution Estimation and Applications
article

A geometric–exponential mixture model for joint inference of discrete and continuous lifetime data

Udo Kamps, Zhe Su
article en

Abstract

Lifetime data from reliability studies are commonly observed continuously over time, but the observations may also be recorded discretely in case of periodic inspections. We study joint statistical inference based on independent lifetime tests with either continuously or discretely observed lifetime data. When the lifetimes of tested items are assumed to be exponentially distributed and if these are only inspected at non-negative integer time points, then the observations correspond to realizations of a discretized version of the lifetime distribution, namely to a geometric distribution with the same distribution parameter involved. We propose and examine a geometric–exponential mixture (GEM) model for the situation, where continuously recorded lifetimes supposed as observations from an exponential distribution as well as lifetimes from identically distributed items monitored at integer time points are available. Statistical inference is then based on all observations. What is most important with the GEM model is that mixtures of a continuous distribution and its discretized version having linked parameters constitute a regular exponential family such that well-known tools and statistical methodology can be utilized. From a minimal canonical representation of the exponential family associated with the GEM model, we obtain the cumulant function, the mean value function, and expressions for divergence measures. We then discuss the existence and uniqueness of maximum likelihood estimators for the distribution parameter and the mixture parameter; moreover, respective uniformly minimum-variance unbiased estimators are considered. Uniformly most powerful (unbiased) tests and uniformly most accurate (unbiased) confidence intervals can be derived. The proposed methods are illustrated by means of two simulation-based data examples. An outlook on a respective family of mixture distributions based on gamma distributions is given.

Statistics
RWTH Aachen University (DE)
Statistical Distribution Estimation and Applications
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon