Optimization of Random Sample Size in Progressively Type II Censoring based on Cost Criterion

Author
Elham Basiri
Kosar University of Bojnord
Abstract
Introduction

‎‎Censored sample arises in a life-testing experiment whenever the experimenter does not observe the failure times of all units placed on a life-test‎. ‎In medical or industrial studies‎, ‎researchers have to treat the censored data because they usually do not have sufficient time to observe the lifetime of all subjects in the study‎. ‎There are different types of censoring‎. ‎The most common censoring schemes are type I and type II censoring schemes‎. Progressively type II censoring is also one of the most important methods of censoring.

One of the most common questions any statistician gets asked is "How large a sample size do I need?"‎. ‎Researchers are often surprised to find out that the answer depends on a number of‎ factors and they have to give the statistician some information before they can get an answer‎. ‎So far different answers have been given to respond this question by considering different criteria‎.

‎Cost criterion is one of the criteria that has always been of interest to researchers‎. ‎So far‎, ‎many researchers have used this criterion for determining the size of samples in different censoring methods‎. ‎

In some applications‎, ‎such as clinical trials and quality control‎, ‎it is almost impossible to have a fixed sample size all the time because some observations may be missing for various reasons‎. ‎In other words‎, ‎the sample size is a random variable‎.

Material and methods

In this paper, a cost function is introduced. Then, assuming that the sample size of progressively type II censoring is a random variable from the truncated binomial distribution, the optimal parameter of sample size distribution in progressively type II censoring, is determined. This optimal parameter is determined so that the introduced cost function does not exceed a pre-determined value, say . In this article, the exponential distribution is considered for lifetimes of observations. A simulation study is also provided to evaluate the obtained results. Finally, the conclusion of the article is presented.

Results and discussion

‎We have computed the values of the expected cost function by considering three different censoring schemes‎. ‎The results show that the expected cost function is an increasing function of m but a decreasing function of θ, ‎when other components are fixed‎, ‎as we expected‎. Also, we can find that considering type II censoring leads to better results than other censoring schemes‎. On the other hand, we can conclude that type II censoring provides the minimum cost among two other censoring schemes. In the sequel, by assuming an upper bound for the cost function, say , the optimal parameter of sample size distribution is obtained.

Conclusion

Determining the optimal sample size is one of the issues that has been studied by many researchers. In some cases, it is not possible for the sample size to be a fixed and pre-determined value. In other words, the sample size is a random variable. In this paper, assuming that the sample size of progressively type II censoring is a random variable from the truncated binomial distribution, the optimal parameter of the sample size distribution is determined. The criterion used in this research is the cost criterion. Next, the optimal parameter of the sample size distribution is determined so that the value of the cost function is less than the specified and predetermined value, say . The results of the paper show that the type II censoring provides less values for the cost function. For all three censorsing schemes, the cost function is an increasing function of m but a decreasing function of θ, ‎when other components are fixed‎, ‎as we expected‎. As a result, the best case scenario is taking into account the type II censoring scheme, selecting smaller values for m, larger values ​​for θ, and smaller values for the parameter of sample size distribution.
Keywords
Random Sample Size
Optimaization
Cost function
‎[1] ‎‎Bain‎, ‎L‎. ‎J‎. ‎and Engelhardt (1991)‎. ‎Statistical Analysis of Reliability and Life-Testing Models‎. ‎Second Edition‎, ‎Marcel Dekker‎, ‎New York‎.

‎‎[2] Cohen‎, ‎A‎. ‎C‎. ‎(1991)‎. Truncated and Censored Samples‎- ‎Theory and Applications‎. ‎Marcel Dekker‎, ‎New York‎.

‎‎[3] Lawless‎, ‎J‎. ‎L‎. ‎(2003)‎. ‎Statistical Models and Methods for Lifetime Data. ‎Second Edition‎, John Wiley and Sons‎, ‎New ‎York.‎

‎‎[4] Nelson‎, ‎W‎. ‎(2004)‎. Applied Life Data Analysis‎. ‎John Wiley and Sons‎, ‎New ‎York.‎

[5] Balakrishnan‎, ‎N‎. ‎and Aggarwala‎, ‎R‎. ‎(2000)‎. Progressive Censoring‎: ‎Theory‎, ‎Methods and Applications‎. ‎Birkhauser‎, ‎Boston‎.

‎[6] Balakrishnan‎, ‎N‎. ‎(2007)‎. ‎Progressive censoring methodology‎: ‎an appraisal‎. ‎Test‎. ‎16‎, ‎211-259‎.

‎[7] Balakrishnan‎, ‎N‎. ‎and Cramer‎, ‎E‎. ‎(2014‎‎‎)‎. The Art of Progressive Censoring‎. ‎Birkhauser‎, ‎New York‎.‎‎

[8] Burkschat, M., Cramer, E., and Kamps, U. (2006). On optimal schemes in progressive censoring. Statistics and probability letters, 76(10), 1032-1036.

[9] Burkschat, M., Cramer, E., and Kamps, U. (2007). Linear estimation of location and scale parameters based on generalized order statistics from generalized Pareto distributions. Recent Developments in Ordered Random Variables. 253-261.

[10] Burkschat, M., Cramer, E., & Kamps, U. (2007). Optimality criteria and optimal schemes in progressive censoring. Communications in Statistics—Theory and Methods. 36(7), 1419-1431.

[11] Volterman W., Davies F.K. and Balakrishnan N. (2012). Pitman closeness as a criterion for the determination of the optimal progressive censoring scheme. Statistical Methodology, 9(6), 563-572.‎

[12] Pradhan, B. and Kundu, D. (2013). Inference and optimal censoring schemes for progressively censored Birnbaum–Saunders distribution. Journal of Statistical Planning and Inference, 143(6), 1098-1108.

[13] Bhattacharya, R., Pradhan, B., and Dewanji, A. (2016). On optimum life-testing plans under Type-II progressive censoring scheme using variable neighborhood search algorithm. Test, 25(2), 309-330.

[14] Mishra, N. (2018). Optimal one-step censoring schemes under entropy criterion. Communications in Statistics-Simulation and Computation. 1-14.

[15] Salemi, U. H., Rezaei, S., Si, Y., and Nadarajah, S. (2018). On Optimal Progressive Censoring Schemes for Normal Distribution. Annals of Data Science, 5(4), 637-658.

‎[16] Pham‎, ‎H‎. ‎(1992)‎, ‎Optimal Design of Life Testing for ULSI Circuit Manufacturing‎, IEEE transactions on semiconductor manufacturing‎, ‎5‎, ‎68-70‎.

[17] Basiri‎, ‎E.‎, ‎and MirMostafaee‎, ‎S‎. ‎M‎. ‎T‎. ‎K‎. ‎(2017)‎, ‎Optimal Sample Size Based on Prediction Problem Under Progressively Type-II Censoring‎, ‎Reliability Theory and its Applications‎, ‎68-72‎.

‎[18] Pham‎, ‎H.‎, ‎and Zhang‎, ‎X‎. ‎(1999)‎, ‎A Software Cost Model with Warranty and Risk Costs‎, IEEE Transactions on Computers‎, ‎48‎, ‎71-75‎.

[19] Cordeiro‎, ‎J‎. ‎B.‎, ‎and Pham‎, ‎H‎. ‎(2017)‎, ‎Optimal Design of Life Testing Cost Model for Type-II Censoring Weibull Distribution Lifetime Units with Respect to Unknown Parameters‎, International Journal of System Assurance Engineering and Management‎, 8‎, ‎28-32‎.

[20] Basiri, E. (2017). Optimal Number of Failures in Type II Censoring for Rayleigh Distribution. Journal of Applied Research on Industrial Engineering, 4(1), 67-74.

‎[21] Raghunandanan‎, ‎K‎. ‎and Patil‎, ‎S‎. ‎A‎. ‎(1972)‎. ‎On order statistics for random sample size‎. Journal of the American Statistical Association‎, 67‎, ‎366-381‎.‎

[22] ‎Buhrman‎, ‎J‎. ‎M‎. ‎(1973)‎. ‎On order statistics when the sample size has a binomial distribution‎. Statistica Neerlandica‎, 27‎, ‎125-126‎.‎‎

[23] ‎Consul‎, ‎P‎. ‎C‎. ‎(1984)‎. ‎On the distribution of order statistics for a random sample size‎. Statistica Neerlandica‎, 38‎, ‎249-256‎.‎‎

[24] ‎Gupta‎, ‎D‎. ‎and Gupta‎, ‎R‎. ‎C‎. ‎(1984)‎, ‎On the distribution of order statistics for a random sample size‎. Statistica Neerlanddica‎, 38‎, ‎13-19‎.‎‎‎‎

[25] ‎Lingappaiah‎, ‎G‎. ‎S‎. ‎(1986)‎, ‎Bayes prediction in exponential life-testing when sample size is a random variable‎. IEEE Transactions on Reliability‎, 35‎, ‎1732-1743‎.‎‎‎‎

[26] ‎Soliman‎, ‎A‎. ‎A‎. ‎(2000)‎, ‎Bayes prediction in a Pareto lifetime model with random sample size‎. ‎Journal of the Royal Statistical Society‎, ‎Series D‎, ‎(The Statistician)‎, 49‎, ‎51-62‎.‎‎‎‎‎

[27] ‎Abd Ellah‎, ‎A‎. ‎H‎. ‎and Sultan‎, ‎K‎. ‎S‎. ‎(2005)‎, ‎Exact Bayesian prediction of exponential lifetime based on fixed and random sample sizes‎. Quality Technology and‎ ‎Quantitative Management‎, 2‎, ‎161-175‎.

[28] ‎Al-Hussaini‎, ‎E‎. ‎K‎. ‎and Al-Awadhi‎, ‎F‎. ‎(2010)‎, ‎Bayes two-sample prediction of generalized order statistics with fixed and random sample size‎. Journal of Statistical Computation and Simulation‎, ‎80‎, ‎13-28‎.

[29] ‎Barakat‎, ‎H‎. ‎M.‎, ‎El-Adll‎, ‎M‎. ‎E‎. ‎and Aly‎, ‎A‎. ‎E‎. ‎(2011)‎, ‎Exact prediction intervals for future exponential lifetime based on random generalized order statistics‎. ‎Computers and Mathematics with Applications‎, ‎61‎, ‎1366-1378‎.

[30] Basiri, E., and Ahmadi, J. (2015). Prediction intervals for generalized-order statistics with random sample size. Journal of Statistical Computation and Simulation. 85(9), 1725-1741.

[31] Basiri, E., and Pakzad, A. (2018). Optimal Non-Parametric Prediction Intervals for Order Statistics with Random Sample Size. Iranian Journal of Management Studies, 11(2), 307-322.

‎[32] Ahmadi‎, ‎J.‎, ‎Basiri‎, ‎E.‎, ‎and MirMostafaee‎, ‎S‎. ‎M‎. ‎T‎. ‎K‎. ‎(2016)‎, ‎Optimal Random Sample Size Based on Bayesian Prediction of Exponential Lifetime and Application to Real Data‎, ‎Journal of the Korean Statistical Society, 45‎, ‎221-237‎.