- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Report Inappropriate Content
Fraction of Design Space Plot
I would like some help making use of the fraction of design space plots to size my DoEs. What I want to be able to do for a given DoE is input my desired precision, the known standard deviation of the response, and the alpha value (confidence level). What I'm expecting then is to get a FDS value in % that describes the percent of the design space that meets my set criteria.
I don't understand what the FDS plot generated in JMP is displaying and where I can set the parameter values myself. Thank you.
Accepted Solutions
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Report Inappropriate Content
Re: Fraction of Design Space Plot
Hi @schlesin,
To clarify the information from FDS plot :
It is generated by calculating relative prediction variance for different design space, and shows the proportion of the design space over which the relative prediction variance lies below a given value : Fraction of Design Space Plot (jmp.com)
It is only linked to the design created and the factor settings, not to the known standard deviation of the response(s) or the confidence level chosen.
You can estimate the actual variance of prediction at any setting by multiplying the relative variance of prediction at that setting with the error variance (mean squared error (MSE) of the model fit for example). You can read the part "Relative Prediction Variance" to get more details : Prediction Variance Profile (jmp.com)
It seems that you may be interested also in Power analysis if you're mentioning confidence level and known standard deviation of your response (and want to check if your design is properly sized, perhaps to be able to screen efficiently significant effects) ?
I hope this answer will help you,
"It is not unusual for a well-designed experiment to analyze itself" (Box, Hunter and Hunter)
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Report Inappropriate Content
Re: Fraction of Design Space Plot
After doing some additional reading, I see that the prediction variance on the FDS plot is a relative prediction variance since the response data is not know. However, it would be nice to know if there is an option to input the known response standard deviation and precision desired so that the design can be properly sized before running the experiment.
In other words, how can I determine whether my design is properly sized?
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Report Inappropriate Content
Re: Fraction of Design Space Plot
Hi @schlesin,
To clarify the information from FDS plot :
It is generated by calculating relative prediction variance for different design space, and shows the proportion of the design space over which the relative prediction variance lies below a given value : Fraction of Design Space Plot (jmp.com)
It is only linked to the design created and the factor settings, not to the known standard deviation of the response(s) or the confidence level chosen.
You can estimate the actual variance of prediction at any setting by multiplying the relative variance of prediction at that setting with the error variance (mean squared error (MSE) of the model fit for example). You can read the part "Relative Prediction Variance" to get more details : Prediction Variance Profile (jmp.com)
It seems that you may be interested also in Power analysis if you're mentioning confidence level and known standard deviation of your response (and want to check if your design is properly sized, perhaps to be able to screen efficiently significant effects) ?
I hope this answer will help you,
"It is not unusual for a well-designed experiment to analyze itself" (Box, Hunter and Hunter)
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Report Inappropriate Content
Re: Fraction of Design Space Plot
Thank you Victor. I will take a look at the power analysis.