This version of the Help is no longer updated. See JMP.com/help for the latest version.

.
Reliability and Survival Methods > Reliability Growth > Statistical Details for the Reliability Growth Platform > Statistical Details for the Piecewise Weibull NHPP Change Point Detection Report
Publication date: 07/30/2020

Statistical Details for the Piecewise Weibull NHPP Change Point Detection Report

The change point is estimated as follows:

Using consecutive event times, disjoint intervals are defined.

Each point within such an interval can be considered to be a change point defining a piecewise Weibull NHPP model with two phases. So long as the two phases defined by that point each consist of at least two events, the algorithm can compute MLEs for the parameters of that model. The log-likelihood for that model can also be computed.

Within each of the disjoint intervals, a constrained optimization routine is used to find a local optimum for the log-likelihood function.

These local optima are compared, and the point corresponding to the largest is chosen as the estimated change point.

Note that this procedure differs from the grid-based approach described in Guo et al. (2010).

Want more information? Have questions? Get answers in the JMP User Community (community.jmp.com).