Optimization method in competing risk mixture for life distribution platformHow does JMP optimize the competing risk mixture in the life distribution platform (what method is used)? Thank you!
AnalyticsOtherRe: Optimization method in competing risk mixture for life distribution platformDo you mean inference method? It is maximum likelihood.
Re: Optimization method in competing risk mixture for life distribution platformIf it's maximum likelihood, how is it calculated? For example, I'm guessing that it's done using the Expectation–maximization (EM) algorithm, because the initial conditions are set using clusters.
Re: Optimization method in competing risk mixture for life distribution platformNo. The optimization method for finding MLE is Newton-Raphson in Life Distribution. E-M could be an alternative for finding MLE in this case, but it is not the only approach.
The likelihood function of a competing risk mixture has the same general form regardless what distribution is involved. The definition of the likelihood in this context can be found in Section 2.4 Likelihood in Statistical Methods for Reliability Data, by Meeker and Escobar.
The likelihood function of a competing risk mixture, in particular, is obtained by replacing the corresponding functions in that section by corresponding functions of the competing risk distribution, whose definition can be found in section 15.2.2 Systems with Components in Series, from the same book. The section defines the distribution function, but survival function and density function can be derived accordingly.
Re: Optimization method in competing risk mixture for life distribution platformThank you for the explanation! If it's Newton-Raphson, then why does the initial conditions have the single cluster, separable cluster, and overlapping cluster setting? Also, if it's on the separable cluster or overlapping cluster setting, then how does JMP decides what part of the data is sectioned off?
