scientific article; zbMATH DE number 1335863
Our focus is on \(k\)-dimensional random vectors with positive coordinate random variables that can be visualized as representing times to failures of \(k\) distinct types. It is envisioned however that an individual will inevitable eventually suffer all \(k\) types of failures so that the coordinates of the random vectors are finite, with probability 1. Mathematically this is a reasonable class of distributions. Additionally, the study of such multivariate survival models has been considered to be a respectable enterprise. It is manifestly distinct from the study of data in which only the identified minimum coordinate is observed (a scenario often described as one involving dependent competing risks). There is, however, a skeleton in the closet. Multivariate survival data are as scarce as hen's teeth. What this actually means is that many multivariate survival models may well find their domain of application in the study of data sets involving \(k\)-dimensional data taking on values in the positive orthant but, realistically, having little to do with survival or times to failure. In a sense they may be considered to be \(k\)-variate distributions motivated by mathematically natural multivariate extensions of univariate survival models. They consequently deserve the name ``multivariate survival distributions even though they relate to data in a manner similar to visible clothes for an invisible emperor (as distinct from the more customary invisible clothes for an all-too-visible emperor). For the entire collection see [Zbl 0913.00046].
This page was built for publication:
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4258552)
