Likelihood and Bayesian estimation of \(P(Y<X)\) using lower record values from a proportional reversed hazard family (Q725670)

A stress-strength model in reliability is meant to describe the life of a component exhibiting a random strength \(X\) and being subject to a random stress \(Y\). The usual measure of component reliability in such a model is given by \(\Phi = P(Y<X).\) This manuscript presents inferencial procedures for \(\Phi,\) when \(X\) and \(Y\) are two independent but not identically distributed random variables and when the information is based on lower record data. The distributions for \(X\) and \(Y\) are assumed to come from the proportional reversed hazard (PRH) family with parameter \(\theta.\) If \(X\) and \(Y\) are PRH distributed with parameters \(\theta_1\) and \(\theta_2,\) respectively, then \(\Phi\) is given by \(\theta_1/(\theta_1+\theta_2).\) The ML estimators and their distributions for \(\theta_1\) and \(\theta_2\) are given. From those, the ML estimator and its distribution for \(\Phi\) are derived. The uniformly minimum variance unbiased estimator of \(\Phi\) is obtained. As for the Bayesian counterpart, the Lindley approximation for the Bayes estimator for \(\Phi,\) as well as its credible set are derived. Numerical optimization techniques provide the HPD interval for such estimator. In particular, they use the Topp-Leone distribution in two real data sets as examples of applications of their theory. The estimation program is largely exploited in this problem. One more nice set of tools is available to be used in reliability.