On the application of compound optimal design strategy in progressively type-II censored life-testing experiments

Abstract

Progressive censoring schemes allow removal of censoring units at the time of each failure. This flexibility of removing items during the experiment makes progressive censoring very effective but it also increases the number of possible censoring schemes. So the problem of choosing optimal progressive censoring schemes has gained a lot of attention in the literature. In this article, we propose a new criterion based on cumulative residual entropy measure for the determination of optimal progressive type II censoring scheme. The proposed criterion represents an overall information of the progressive type II censored life-testing experiment and we obtain optimal schemes by maximizing this measure. Also, we develop optimal design by maximizing the information subject to a given cost constraint. Finally, we perform a compound optimal design strategy that simultaneously optimizes our proposed criterion and the total cost of the experiment. Our proposed design does not depend on any asymptotic results. Sensitivity analysis for the constraint design and COD are studied for the mis-specification of the input parametrs. A real data set is analyzed for the purpose of illustration.