cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
Browse apps to extend the software in the new JMP Marketplace
Choose Language Hide Translation Bar
AgnieszkaL
Level II

Curve DoE - defect rate issue?

Hi All

 

I’ve just started using Curve DoE function and have came across the following challenge:

The defect rate is ~70% - not sure how does this relate to the entire curve? Or just a selected point on the curve? Also, I need max desirability on all response values across the curve concertation range hence the desirability curve is flat – not sure though if that’s OK? Could you comment pls?

AgnieszkaL_1-1715938047724.png

 

1 ACCEPTED SOLUTION

Accepted Solutions
SamGardner
Level VII

Re: Curve DoE - defect rate issue?

The issue here is that the Curve DOE model is predicting the curve, which is shown on the left panel in the profiler.

If the curve value at a given time point is something that you want to achieve or optimize, the best approach would be to fix the X value for the curve (lock it by Cntl-Click on the graph and select the "Lock Factor Setting").  Then do not use a random distribution for inputs for that X factor.

 

You can use the "Maximize For Each Grid Point" option to sequence the locked factor over 5 steps (from the min to the max of the X in the graph).   

View solution in original post

3 REPLIES 3

Re: Curve DoE - defect rate issue?

The defect rate is based on the simulated values across the whole of the curve, as you have set all of your values as 'Random Normal' - 

 it may be useful to set your X values for your curve (which looks to be your left most plot) to 'Random Uniform', this way you get provided a defect rate across your whole range. Another option is to keep it 'Fixed' to provide a defect rate for a specific point within the curve. 

“All models are wrong, but some are useful”
SamGardner
Level VII

Re: Curve DoE - defect rate issue?

The issue here is that the Curve DOE model is predicting the curve, which is shown on the left panel in the profiler.

If the curve value at a given time point is something that you want to achieve or optimize, the best approach would be to fix the X value for the curve (lock it by Cntl-Click on the graph and select the "Lock Factor Setting").  Then do not use a random distribution for inputs for that X factor.

 

You can use the "Maximize For Each Grid Point" option to sequence the locked factor over 5 steps (from the min to the max of the X in the graph).   

AgnieszkaL
Level II

Re: Curve DoE - defect rate issue?

Thanks a lot - that is useful to know!