Masked Data Analysis based on the Generalized Linear Model

Document Type : Original Article


School of Mathematics, Statistics and Computer Science, College of Science, University of Tehran, Tehran, Iran


In this paper, we consider the estimation problem in the presence of masked data for series systems. A missing indicator is proposed to describe masked set of each failure time.Moreover, a Generalized Linear model (GLM) with appropriate link function is used to model masked indicator in order to involve masked information into likelihood function. Both maximum likelihood and Bayesian methods were considered.The likelihood function with both missing at random (MAR) and missing not at random (MNAR) mechanismsare derived.Using an auxiliary variable, a Bayesian approach is expanded to obtain posterior estimations of the model parameters.The proposed methods have been illustrated through a real example.


Main Subjects

[1]      Miyakawa, M. “Analysis of incomplete data in a competing risks model,” IEEE Trans. Rel. 33(4): pp. 293–296, 1984. 
[2]      Basu, S. Basu, A. P. and Mukhopadhyay, C. “Bayesian analysis for masked system failuredata using nonidenticalweibull models,” J. Statist. Plann. Inference, vol. 78, pp. 255–275, 1999.
[3]      AziziF. and Haghighi,F. Joint modeling of linear degradation and failure time data with masked causes of failure under simple step-stress test.  J Stat Comput Simul.  88(8):1603–1615, 2018. doi: 10.1080/00949655.2018.1442468.
[4]      Usher, J. S. and Hodgson, T. J.  “Maximum likelihood analysis of component reliability usingmasked system life data,” IEEE Trans. Rel., vol. 37, no. 5, pp. 550–555, 1998.
[5]      Lin, D. K. J. Usher, J. S. and Guess, F. M. “Exact maximum likelihood estimation usingmasked system data”, IEEE Trans. Rel., vol. 42, no. 4, pp. 631–635, 1993.
[6]      Reiser, B. Guttman, I. Lin, D. K. J. Usher, J. S. and Guess, F. M. “Bayesian inference formasked system lifetime data,” Appl. Statist., vol. 44, pp. 79–90, 1995.
[7]      Berger, J. O. and Sun, D. “Bayesian analysis for the poly-Weibull distribution,” J. Amer.Statist. Assoc., vol. 88, pp. 1412–1418, 1993.
[8]      Mukhopadhyay, C.  and Basu, A.  P.  “Bayesian analysis of incomplete time and cause offailure data,” J. Statist. Plann. Inference, vol. 59, pp. 79–100, 1997.
[9]      Cai, J., Shi, Y., Zhang, Y. Robust Bayesian analysis for parallel system with masked data under inverse Weibull lifetime distribution. Commun Stat-Theory Methods. 49:1–13, 2019.
[10]  Sen, A.  Banerjee, M.  and Basu, S.  Balakrishnan, N.  and Rao, C.  R.  Eds.  “Analysis ofmasked failure data under competing risks,” in Handbook of Statistics. Amsterdam, The Nether-lands:  North-Holland, vol. 20, pp. 523–540, 2001.
[11]  Basu, S.  Sen, A.  and Banerjee, M.  “Bayesian analysis of competing risks with partiallymasked cause of failure,” Appl. Statist., vol. 52, pp. 77–93, 2003.
[12]  Lin D. K. J. and Guess, F. M. “System life data analysis with dependent partial knowledgeon the exact cause of system failure”, Microelectron. Rel., vol. 34, pp. 535–544, 1994.
[13]  Guttman, I.  Lin, D.  K.  J.  Reiser, B.  and Usher, J.  S. “Dependent masking and systemlife data analysis:  Bayesian inference for two-component systems,” Lifetime Data Anal., vol. 1, pp.87–100, 1995.
[14]  Kuo, L. and Yang, T. E. “Bayesian reliability modeling for masked system lifetime data”,Statist. Probab. Lett., vol. 47, pp. 229–241, 2000.
[15]  Mukhopadhyay, C. and Basu, A. P. “Masking without the symmetry assumption: A Bayesianapproach,” in Proc. Abstract Book 2nd Int. Conf. Math. Methods Rel., Bordeaux, France, vol. 2,pp. 784–787, 2000,Universite Victor Segalen.
[16]  Craiu, R. V. and Duchesne, T. Gelman, A. and. Meng, Eds. X. L. “Using EM and DA for thecompeting risk model,” in Applied Bayesian Modeling and Causal Inference from an Incomplete-Data Perspective. New York, NY, USA: Wiley, pp. 234–245, 2004.
[17]   [17] Mukhopadhyay, C. “Maximum likelihood analysis of masked series system lifetime data,” J.Statist. Plann. Inference, vol. 136, pp. 803–838, 2006.
[18]  Xu, A. and Tang, Y. “Bayesian analysis of Pareto reliability with dependent masked data,”IEEE Trans. Rel., vol. 58, no. 4, pp. 583–588, 2009.
[19]  Xu, A. Basu, S. and Tang, Y.” A Full Bayesian Approach for Masked Data in Step-StressAccelerated Life Testing,” IEEE Trans. Rel., vol. 63, no. 3, pp. 798-806, 2014.