I tried to comment on my wish list item to add additional context (and link to a relevant JMP User Community post), however it was marked as spam.
Link to post:
(Life Distribution plot points default to Kaplan-Meier or Turnbull estimates with option to select midpoints)
by TomH
https://community.jmp.com/t5/JMP-Wish-List/Life-Distribution-plot-points-default-to-Kaplan-Meier-or/idi-p/718233
What inspired this wish list request? As a reliability engineer, I frequently use the Life Distribution platform for its distribution plotting and fitting capabilities. These plots are important not only for internal topics but also for critical customer communications. It appears that JMP 17 has changed the method for plotting data points in this platform in a way that is less desirable. I have not been able to find any notification of this change on the JMP website, e.g. New in JMP 17, which is rather disappointing. These plots are often used to demonstrate the quality of fit of a distribution model to data. Such demonstrations can be crucial in obtaining acceptance. The Kaplan-Meier / Turnbull estimates used in JMP 16 were appropriate for this, as a good MLE distribution fit often came close to the estimates. For example, see the Weibull MLE fits below (same data table, different versions). Both MLE Weibull fits are identical, however JMP 17 points are significantly below the fit. This change to the default behavior is troubling - now I must question which version of JMP was used when reviewing plots. JMP 16 – “great fit!” JMP 17 – “not such a great fit?” (same fit!) In JMP 17 the user can select Points (black, default) / Step Function (red) / Both / None. Both selected below: However there are a some issues with this: User has no customization options for the red markers (which are what should be default). Need ability to adjust color, marker, size, etc. just like regular markers. Buggy behavior: selecting Step Function does not work when going from one of the other options. It works when the platform is first launched if selected as default in preferences, however when alternating between Points and back to Step Function it once again does nothing. Even if selected as default in preferences, once the “Nonparametric” distribution is deselected the red markers disappear and can only be recalled if “Both” selected. What is the improvement you would like to see? The default point positions should be the “step function” points (Kaplan-Meier / Turnbull estimates) to be consistent with previous versions. It would still be nice to be able to select midpoints, however each option should have full marker customizability. Why is this idea important? These plots are often used to demonstrate the quality of fit of a distribution model to data. Such demonstrations can be crucial in obtaining acceptance. The previous default Kaplan-Meier / Turnbull estimates were appropriate for this, as a good MLE distribution fit often came close to the estimates. Deviating from this now with JMP 17 is a misstep that should be corrected. JMP® Pro 17.2.0 (701896) Microsoft Windows 11 Enterprise (10.0.22621.0)
This message has 0 replies