21973_en-US

liliconcom

Random Seed Reset

Run this add-in to set the random seed to a specified value.

 

Notes/Disclaimers:

 

• The add-in can be used to set the random seed prior to creating a validation column, prior to simulating values (using the formula editor or the Simulator), and in many analysis platforms (run prior to launching the platform).
• If using the DOE Custom Designer, use the built in Set Random Seed option.
• If using Partial Least Squares in JMP Pro, use the built in Set Random Seed option for Kfold and Holdback validation.
• Automatic cross-validation using a validation column is only available in JMP Pro.
• The random seed does not persist across analyses (the add-in must be re-run each time using the same random seed).
• Test when using with any analysis platform to verify that you’re seeing the same results.
• Dec 7, 2017: There is a known problem in JMP 13 that can result in different results across different machines and operating systems.

 

Updated 6/4 (streamlined the use notes).

Updated 7/29 (minor code changes - added title to Panel Box() call and Names Default to Here(1))

 

Developed for academic/teaching purposes with BradyBrady and susan.walsh1, with jordanhiller@jmp .

Predictive Modeling and Data MiningRe: Random Seed Reset

Does the issue with differing results still occur in JMP 14? We have been using JMP 13 to teach a course. Students get different answers using a Mac or PC (or different versions of JMP 13) which is very inconvenient for homeworks.

Re: Random Seed Reset

RE: the Dec 7, 2017 comment
I believe the differing results only occur with a very small minority of platforms.

Re: Random Seed Reset<meta http-equiv="Content-Type" content="text/html; charset=UTF-8" />

A revised version of this add-in has been posted (7/29/2015), with some minor code adjustments (noted above).  Thanks to Sam Gardner for identifying these improvements.

Re: Random Seed Reset

Just a reminder, per a conversation with Mia Stephens:  The Add-In works in JMP Pro and JMP.  In JMP ,the user would have to manually apply the validation column using Data Filter or Hide and Exclude.