Our World Statistics Day conversations have been a great reminder of how much statistics can inform our lives. Do you have an example of how statistics has made a difference in your life? Share your story with the Community!
gail_massari
Community Manager Community Manager

Reliability for Non-Repairable Systems - Journal and Data

Data and scripts to use to practice reliability analysis using JMP.  View the .video,

 

If interested, see attached Genschel and Meeker paper, A Comparison of Maximum Likelihood and Median-Rank Regression for Weibull Estimation.

Comments

Thank you for this contribution! @JerryFish  and @gail_massari . One of the biggest hurdles in my mind when being new to reliability methods is just getting familiar with the data structure, which is obviously a little bit different than your typical experimental or retrospective data analysis scenario (without any component of time dependence).  

 

The example with the competing risk mixture was particularly interesting to me as it relates to my work. Many times, there will be some distributional evidence of more than one failure mode in testing for bond breaks for example (the Normal-Quantile plot is a good tool for detecting this in the Distribution Platform), but we won't necessarily know which failure mode is which, since, unlike in the analysis case  @JerryFish described where the technician was "holding out on us" at first, then gave us the correct failure causes, our technician may not "diagnose" the appropriate failure mode correctly (because it may be difficult to consistently diagnose).   In such cases, do you know if your customers have had any success fitting a competing risk mixture, then using that fit separation as the basis for separating that data and analyzing it separately by fit in that way? (e.g. performing Cpk analysis to estimate product quality (on "Fit 1" and "Fit 2" assuming a Mixture of 2 distributions)  @JerryFish brought up a good point about the importance of relying on subject-matter-expertise to make a decision about the appropriate distributional model to fit; often times it's a delicate push-pull between what we can find empirically using the power of JMP (AIC, BIC, -2*Logliklihood) and what we expect to see based on the physics/engineering.   -@PatrickGiuliano 

Article Labels
Article Tags