Predicting Failure times for some Unobserved Events with Application to Real-Life Data

TL;DR

This paper develops a statistical method to predict failure times of unobserved units in lifetime experiments with multiply-hybrid censored data, using linear failure rate distribution and simulation techniques for real-life applications.

Contribution

It introduces a novel two-sample prediction approach for failure time estimation under multiply-hybrid censored data using linear failure rate distribution.

Findings

Linear failure rate distribution fits real-life data well

The proposed prediction method effectively estimates failure times

Simulation validates the approach's accuracy

Abstract

This study aims to predict failure times for some units in some lifetime experiments. In some practical situations, the experimenter may not be able to register the failure times of all units during the experiment. Recently, this situation can be described by a new type of censored data called multiply-hybrid censored data. In this paper, the linear failure rate distribution is well-fitted to some real-life data and hence some statistical inference approaches are applied to estimate the distribution parameters. A two-sample prediction approach applied to extrapolate a new sample simulates the observed data for predicting the failure times for the unobserved units.