What inspired this wish list request? Please describe the current issue that needs improvement or the problem to be solved that is not easy or possible right now, with an example use case. The variable importance analysis employs a Monte Carlo simulation, therefore each time you perform a specific analysis the importance report changes somewhat, and the results are therefore not exactly reproducible. One can manually add a random seed command to the Profiler script to make the importance report reproducible in the future. However, it would be nice to be able to set a random seed while performing the analysess.
What is the improvement you would like to see? Please describe the idea for improving JMP. Please include mock-ups, wireframes, screenshots, scripts, other documents or examples from other software that help describe the change you would like to see.
Exact reproducibility of variable importance results
Why is this idea important? Please describe the value to you and/or other users if the idea is implemented (for example, ease of use, must have,…).
When scientists observe that the variable importance reports are not exactly the same each time the analysis is performed, they tend to become skeptical of the approach. Also, these variable importance reports are sometimes used as supporting documentation for critical process parameter identification in FDA filings, so it is important they be exactly reproducible.